processing of visual information in dim light. functional variability...

1 0 0 m s

2 m V

100 ms10 mV

KYÖSTI HEIMONEN

REPORT SERIES IN PHYSICAL SCIENCES

Report No. 52 (2008)

PROCESSING OF VISUAL INFORMATION IN DIM LIGHT

Functional variability, matched filtering and spike coding

in cockroach (Periplaneta americana) photoreceptors

KYÖSTI HEIMONEN Department of Physical Sciences Division of Biophysics and Biocenter Oulu University of Oulu Finland

Academic dissertation to be presented, with the permission of the Faculty of Science of the University of Oulu, for public discussion in the Auditorium GO101, Linnanmaa, on December 5th, 2008, at 12 o’clock noon.

REPORT SERIES IN PHYSICAL SCIENCES Report No. 52

OULU 2008 UNIVERSITY OF OULU

PROCESSING OF VISUAL INFORMATION IN DIM LIGHT

Functional variability, matched filtering and spike coding

in cockroach (Periplaneta americana) photoreceptors

2

Opponent Prof. Eric J. Warrant, Lund University, Sweden Reviewers Prof. Andrew S. French, Dalhousie University, Canada Prof. Eric J. Warrant, Lund University, Sweden Custos Prof. Matti Weckström, University of Oulu, Finland ISBN 978-951-42-8948-4 ISBN 978-951-42-8949-1 (PDF) ISSN 1239-4327 Oulu University Press Oulu 2008

3

“Nothing changes but the colour changes hue” “Nothing changes but the channel changes view”

- Midnight Oil (2001) Too Much Sunshine. In: Capricornia -

5

Heimonen, Kyösti: Processing of visual information in dim light. Functional variability, matched filtering and spike coding in cockroach (Periplaneta americana) photoreceptors. Faculty of Science, Department of Physical Sciences, Division of Biophysics, University of Oulu, P.O. Box 3000, FI-90014 University of Oulu, Finland; Biocenter Oulu, University of Oulu, P.O. Box 5000, FI-90014 University of Oulu, Finland. Report Series in Physical Sciences No. 52 (2008).

Abstract

Sensory systems are considered to be optimized for their ecological niche. In vision this means highly organised regular structure and function, where nearly identical photoreceptors have graded light responses in order to be able to handle as much information as possible. Instead, cockroach compound eyes show large amounts of irregularities in their optics and structure, and unusually long axons.

In this thesis photoreceptors of the cockroach were studied with intracellular recordings of their light responses, biophysical systems analysis, and modelling of the relations between the light stimuli and responses. Cockroaches prefer living in dark or extremely dim environments. However, they have large and complex compound eyes. The aim of this study was to find out the functional properties by which the visual system and especially photoreceptors have adapted to cope with, i.e. to see in, dim light conditions.

The function of photoreceptors was found to vary randomly in many respects, and the long axons seemed to utilise action potential coding of visual signals. Through model simulations it was shown that signals of a group of these functionally variable and spiking photoreceptors, when pooled, could provide more reliable coding than signals of identical cells of any experimentally characterised type. This naturally sacrifices spatial resolution. The filtering dynamics of the photoreceptors is matched to low light intensities and their temporal resolution does not markedly improve with increasing light adaptation. Adaptation processes in the photoreceptors saturated near an intensity of about 1000 effective photons/s. These are all both unexpected and novel features of photoreceptor function. Spatial summation of functionally different photoreceptors and reduced temporal resolution and contrast coding abilities can be considered to be permanent optimizations to a dim environment.

Keywords: vision, sensory coding, adaptation, temporal, spatial, resolution.

7

Acknowledgements

This thesis has been done at the Department of Physical Sciences, Division of Biophysics in the University of Oulu. I owe my highest appreciation and deepest gratitude to my supervisor, and dear friend, Professor Matti Weckström, the head of the division and biophysics research group. I want to thank him, from the bottom of my heart, for the opportunity to do this thesis and the excellent guidance, support, ideas, discussions, and collaboration during the whole process. Without him this work would still be undone. I also wish to thank all the professors, especially Jukka Jokisaari, the head of the department, and everybody else at the department for creating a great scientifically oriented atmosphere. Working at the department has been a privilege.

I’m thankful to everybody at biophysics for their collaboration and friendship, and all the brilliant discussions. Especially, I want to thank Esa Luoma for his help with the technical and computer problems, Iikka Salmela for his help with the recordings and figures, Panu Kontiokari for programming and running the simulations, Arto Piironen for programming the Hilbert transform and Mikko Vähäsöyrinki for his competent help with various scientific problems and tasks. I also want to thank Mikko Lempeä for his expertise on building several of the “gadgets” crucial for the experiments and Anja Mäntykenttä for her help with the practical arrangements.

Some early parts of the thesis were done at the Department of Physiology and the Department of Biology in the University of Oulu. I’m also grateful to everybody involved with my work there. Particularly, I want to thank Eero Kouvalainen for his friendship and exquisitely competent technical assistance, Mikko Juusola for programming the recording and analysis programs and Pasi Tavi for his friendship and expert comments on diverse scientific issues.

I also wish to express my appreciation and regard to everybody who participated in commenting, correcting and improving the manuscript and its language, but especially to Professors Eric Warrant and Andrew French.

Finally, I wish to express my ultimate gratitude to Leena Inkala for showing me that there are also other things in life (apart from science) worth experiencing. Her support, comfort, thoughts and love have helped me to bear myself during this process. Katajaranta, November 2008 Kyösti Heimonen

9

Abbreviations

A Amplitude (mV) of a light response Amax Maximum amplitude of the light response c Contrast (=ΔI/I) Cm Capacitance of the cell membrane DC Direct current gm Conductance of the cell membrane ΔI Change in the light intensity Δρ Acceptance angle, i.e. the width of the photoreceptor receptive field at 50% sensitivity level Erev Reversal potential of the current through the cell membrane ERG Electroretinogram FFT Fast Fourier transform I Intensity of light Im Current through the cell membrane ISI Inter-stimulus interval LED Light emitting diode N Number of photons n 1) Usually number of experiments made or numerical values used to calculate an average value 2) In the special case of the modified (by Laughlin 1981) equation of Lipetz (1971) the exponent that is added to consider the effects of additional sources of nonlinearity R Reciprocal of the light intensity (I) producing a half-maximal (Vmax/2) amplitude for the light response Rm Resistance of the cell membrane SD Standard deviation SNR Signal-to-noise ratio, i.e. the power or amplitude of the signal divided with the corresponding value of the noise τm Time constant of the cell membrane UV Ultraviolet V-logI curve A curve, where the maximum amplitude of the photoreceptor voltage response (Vmax) is plotted in function of the stimulus light intensity (I) Vm Voltage over the cell membrane Vmax Maximum amplitude of the light response

11

List of original articles

This thesis is based on the following papers, which are referred to in the text by their Roman numerals I – III:

I Heimonen K, Salmela I, Kontiokari P & Weckström M (2006) Large functional variability in cockroach photoreceptors: An optimization to low light levels. J Neurosci 26: 13454-13462 (+ supplementary material).

II Heimonen K, Juusola M & Weckström M (2008) Are the linear dynamics of

cockroach photoreceptor responses matched to low levels of luminance? Manuscript.

III Weckström M, Järvilehto M & Heimonen K (1993) Spike-like potentials in the axons of nonspiking photoreceptors. J Neurophysiol 69: 293-296.

In the first two articles (I, II) the measurements, their analysis and writing were mainly done by Heimonen. Weckström had crucial influence at the level of ideas and all the authors participated in writing. In paper I Salmela and Weckström helped with the recordings and Kontiokari was primarily responsible on the programming of the simulation. Juusola programmed the original recording and analysis programs in paper II. The third article (III) is partly based on the recordings and their analysis in the master thesis of Heimonen (1990) supervised by Järvilehto. Additional recordings, modelling and writing of the first version were done by Weckström.

13

Contents

Abstract 5 Acknowledgements 7 Abbreviations 9 List of original articles 11 Contents 13 1 Introduction 15 2 Review of the literature 18

2.1 Processing of visual information in photoreceptors ................................ 18 2.1.1 Spatial resolution and spectral sensitivity .................................... 19 2.1.2 Phototransduction and graded light responses.............................. 20 2.1.3 Adaptation .................................................................................... 22 2.1.4 Temporal resolution...................................................................... 23

2.2 Vision in dim light................................................................................... 26 2.2.1 Noise............................................................................................. 26 2.2.2 Temporal and spatial summation.................................................. 27

2.3 Vision of the cockroach........................................................................... 28 2.3.1 Visually guided behaviour of the cockroach ................................ 30 2.3.2 Structure of the cockroach visual system ..................................... 33 2.3.3 Function of the cockroach visual system...................................... 38

3 Aims of the research 45 4 Summary of the material and methods 46

4.1 Animals and preparation ......................................................................... 46 4.2 Recording and stimulation equipment and procedures ........................... 48 4.3 Quality of recordings .............................................................................. 50 4.4 Marking and anatomical identification of cells with spike

responses ................................................................................................. 50 4.5 Localization of recorded photoreceptor responses.................................. 51 4.6 Recordings of photoreceptor impulse and step responses in dark

adaptation................................................................................................ 51 4.7 Recordings of photoreceptor step responses in light adaptation ............. 53 4.8 Determination of receptive field ............................................................. 53 4.9 Determination of signal-to-noise ratio .................................................... 53 4.10 Determination of frequency response ..................................................... 54 4.11 Evaluation of quantum bump size and shape in light adaptation ............ 55 4.12 Determination of impedance function..................................................... 55

14

4.13 Statistical tests ......................................................................................... 55 4.14 Modelling and simulations ...................................................................... 56

5 Summary of the results in the original articles 57 5.1 The spatial filter ...................................................................................... 58 5.2 Properties of responses in dark and light adaptation and their

variability ................................................................................................ 59 5.3 The SNR of contrast coding .................................................................... 62 5.4 The frequency response........................................................................... 63 5.5 The transduction filter ............................................................................. 64 5.6 The membrane filter ................................................................................ 67 5.7 The “axon filter” and spike coding ......................................................... 68 5.8 A hypothesis concerning the function of the cockroach visual

system...................................................................................................... 69 6 Discussion 70

6.1 Functional variability of the photoreceptors ........................................... 70 6.2 Is functional variability of photoreceptors optimization to low

light levels? ............................................................................................. 74 6.3 Is filtering of visual information in cockroach photoreceptors

matched to low levels of luminance? ...................................................... 75 6.4 Is the visual information spike coded in the cockroach

photoreceptor axons? .............................................................................. 76 6.5 Are cockroach photoreceptors optimized for vision in dim light? .......... 79

7 Conclusions 81 References 82 Original articles 97

15

1 Introduction

Information processing in neurons or neuronal systems is based on electrical signals produced over and travelling along the membranes of neuronal cells and chemical (synapses) or electrical (gap junctions) communication between the neurons. The biophysical and biochemical basis and the physiology in general behind these processes can be found in many textbooks on neuroscience, biophysics or physiology (e.g. Johnston & Wu 1995, Koch 1999, Kandel et al. 2000, Hille 2001, Dayan & Abbott 2005). At the level of a neuron this information processing includes several complex cell physiological and biophysical phenomena, for example:

1. Homeostatic control of the cellular environment, e.g. uneven distribution of the ionic constituents of intra- and extracellular fluids and their transportation through the membrane, control of Nernst and resting potentials etc.

2. Generation of electrical signals with the ionic currents through the membrane via controlled but stochastic closing and opening of the membrane-bound ion channels.

3. Coding of the carried information into electrical signals of the cell membrane, i.e. graded responses (continuous, or analogue, signals) and/or action potentials (pulse coded or “digitized” signals).

4. Propagation of the signals along the neuronal cable formed by the membrane.

Consequently the production and propagation of the electrical signals carrying the necessary information are controlled and their changes actively processed by the neurons themselves. The roles of the other major cell groups of nervous systems, collectively called "glial cells" are not completely understood, and are generally considered to be supportive of neurons. However, they may also be found to contribute to information processing.

In a sensory neuron the information is harvested from the physical or chemical nature of the surroundings, not from the signals of other neurons, and transformed into electrical signals of the cell membrane via the transduction process (reviewed e.g. by French 1992, Gold 1999, Herness & Gilbertson 1999, Burns & Baylor 2001, Hardie 2001, Hardie & Raghu 2001, Bhandawat et al. 2005, Hardie 2006, Lumpkin & Caterina 2007, Vollrath et al. 2007). In a sensory transduction process the energy released by the absorption of an adequate stimulus (e.g. a photon) is absorbed by a membrane bound specialized receptor molecule (e.g. a rhodopsin photopigment). This triggers a biochemical chain of

16

reactions called the transduction cascade, which ultimately leads to opening of ion channels and generation of electrical signals.

The interest to study especially visual systems is partly based on the fact that for us humans vision is such an essential and central sensory system. The second good reason to use vision as a target of studies in sensory and neural processing is more pragmatic. Our ability to control visual inputs, light stimuli, down to their finest details (direction, wavelength, intensity etc.) makes vision a very benign and beneficial sensory system from the point of view of biophysical systems analysis. This ability enables the exact control and recording of both the stimulus (input) and the response (output). Thirdly, many photoreceptors are fairly large in size, especially in diameter (to capture light effectively), when compared to most other sensory cells, and thus accessible to intracellular recording.

In visual systems, here particularly invertebrate photoreceptors, processing of visual information and membrane responses generated by the visual neurons have been and are actively studied (e.g. books and reviews: Autrum 1958, Wolken 1971, Järvilehto 1978, Autrum 1981, Land 1981, Laughlin 1981, Shaw 1984, Weckström 1987, Laughlin 1989, Stavenga & Hardie 1989, Warrant & McIntyre 1992, Juusola 1993a, Uusitalo 1995, Weckström & Laughlin 1995, Juusola et al. 1996, Land 1997, Hardie & Raghu 2001, Vähäsöyrinki 2004, Warrant & Nilsson 2006, Frederiksen 2008). Thus processing of visual information is well known in several invertebrate species, especially in the blowfly Calliphora, the horseshoe crab Limulus and lately also in the fruitfly Drosophila.

In the course of animal evolution vision has played a key role in sensing the physical world (Parker 2004). Accordingly, visual systems have evolved and can be considered to be well adapted, or even optimized, to their environments and tasks (Laughlin 1989, Laughlin 1990, Weckström & Laughlin 1995, Laughlin 1996, Warrant 1999, Warrant 2004). This has produced many extraordinary and even peculiar solutions for the structure and function of the visual system. Although darkness literally means something one can’t see, some animals have not lost their eyes while evolving to live in and cope with dim, and to us humans even dark, conditions. Instead, they have even retained fairly complex visual systems and visually guided behavioural tasks (e.g. Kelber et al. 2002, Dacke et al. 2003, Dacke et al. 2004, Warrant et al. 2004, Warrant 2008). A fair amount is known about seeing in dim light (reviewed e.g. by Hess et al. 1990, Laughlin 1990, Warrant 1999, Warrant 2004, Warrant 2006, Frederiksen 2008), but also a great deal remains a mystery.

17

This thesis is based on papers (I, II, III), where the visual system and primarily the photoreceptor cells of the American cockroach (Periplaneta americana) have been studied. This was done through intracellular recordings of the electrical signals of the cell membranes responding to controlled light stimuli and via biophysical systems analysis of the relations between the stimuli and the responses. Biophysical modelling of the function of cockroach photoreceptors and visual system was also utilized. Cockroaches prefer living in dark or extremely dim environments (e.g. Guthrie & Tindall 1968, Bell & Adiyodi 1982, Huber et al. 1990a, Huber et al. 1990b, Mote 1990, Halloy et al. 2007). However, they have retained large and complex compound eyes in the course of evolution (Butler 1973a, Butler 1973b, Mote 1990). Thus the emphasis of this study has been placed on revealing the functional features by which their visual system, and especially their photoreceptors, have adapted to cope with, i.e. to see in, dim light conditions.

18

2 Review of the literature

The information brought into the eye by the incident light is in the form of spatial, spectral, intensity and polarisation information. In other words the light carries information about the direction of its arrival, the colour (wavelength or energy of photons) and the brightness (number of photons) of its source and possibly about the angle of the plane it is polarized in. In addition, all this information can change temporally, i.e. the time of arrival can vary, through self-motion, temporal intensity variation in the surroundings, etc. The processing of this information takes place in the visual system, where the photoreceptor cells with the optical structure of the eye form the sensory apparatus. Hence, it is the photoreceptors (with the aid of the optics) who dictate, which features and parts of the available light information are going to be captured and transduced into the electrical signals for the whole neural network of vision to be further processed.

Since this thesis deals with the photoreceptors of an insect, the American cockroach (Periplaneta americana), the approach taken here is based mainly, if not almost completely, on the literature on the insect and compound eye photoreceptors. Even this limited field of vision research is so widely studied that this review had to be concentrated almost solely on the subjects and viewpoints covered in the original articles of this thesis (I, II, III). Only what little is known about the vision and the visual system of the cockroach is reviewed here more extensively (section 2.3). It is to be noted that there is a huge amount of literature on the subject of visual information processing, especially when considering also the vision of vertebrates, which is completely omitted here.

2.1 Processing of visual information in photoreceptors

The word “processing” represents here all the modifications the photoreceptor executes on the information arriving with the absorbed photons. At first, light information is absorbed and transduced into information that is in the form of a changing membrane potential i.e. the light response signals. Then these electrical membrane signals are carried or propagated along the neuronal cable to the first synapse. All the processing the photoreceptor performs alters or filters the information contents of the signals. A “filter” or “filtering” is a widely used way to describe a certain feature of this processing. It gives a relative measure of membrane potential amplitude or some other form of scaled strength or power of

19

the response as a function of a certain quality of the stimulus, e.g. direction, colour, speed, frequency etc.

2.1.1 Spatial resolution and spectral sensitivity

Before the biochemical and electrical events in the photoreceptors take place the optics of the eye process the light information. The main purpose is to focus the desired portion of the incident light (or photons) into the light absorbing or sensitive structures of the photoreceptors. The compound eyes are formed of “eye-units” or “partial eyes” called ommatidia, whose number varies from a few to tens of thousands between animal species and developmental stages (reviews e.g. Wolken 1971, Land 1985). Each ommatidium has its own optical system and photoreceptors behind it. Several extensive reviews and articles are available (e.g. Kirschfeld 1971, Land 1981, Land 1985, Nilsson 1990, Land 1997, Frederiksen 2008) on the optical arrangement of the different compound eyes (apposition, neural superposition and diverse optical superposition eyes) and their function. They are not reviewed any further here, only the structure and optics of the cockroach apposition eye is covered in more detail later (section 2.3.2).

Basically the optical axes of the ommatidia in a compound eye each have their own slightly different directions and thus the photoreceptors in them have their own visual fields (e.g. Land 1981, Land 1997). In a fused rhabdom, like in the cockroach (Snyder & Horridge 1972, Snyder et al. 1973), all the photoreceptors inside one ommatidium share the same field of vision. Roughly speaking it can be thought that there is a mosaic picture formed on the retina of the compound eye (Kirschfeld 1971), where the grain or “pixel” size is defined by the visual fields of the ommatidia. Hence, it is the optical arrangement of the compound eye that mainly forms and dictates the visual field or spatial filter of a photoreceptor (e.g. Land 1981, Land 1997). The spatial filter is usually presented in the form of a receptive field or angular sensitivity function, where the scaled or normalized amplitude of the light response, i.e. photoreceptor sensitivity, is plotted as a function of the angle of the stimulating light (see Fig. 4 in I). The angular sensitivity functions are normally considered to have a Gaussian shape, because of diffraction in a tiny lens. Thus they can be described by a single number, the acceptance angle Δρ, usually defined as the width (in degrees) of the Gaussian at 50% sensitivity level.

The optical properties and spatial filtering can not be totally separated from all the processing of light information taking place in the photoreceptors. To some

20

extent the optical properties of the ommatidia are dependent on the photoreceptors in them, namely the properties of rhabdomeres and pigments in the cells (e.g. Snyder & Horridge 1972, Snyder et al. 1973, Stavenga 2003a, Stavenga 2003b, Stavenga 2004). Partially the optics regulates, restricts and modifies the function of photoreceptors, for instance the spectral contents of the light reaching the photoreceptors can be (and is) altered.

The next filter that shapes the information in incident light is formed by the internal properties of photoreceptors themselves. The spectral sensitivity of a photoreceptor is principally defined by the properties of the photopigment molecules embedded in the rhabdomeric membrane (e.g. Stavenga & Schwemer 1984, Hardie 1986, Stavenga & Hardie 1989). These visual pigments absorb the incident photons and start the phototranduction cascade. The probability of the absorption is dependent on the wavelength or energy of the incident light. This can be presented in the form of a spectral sensitivity function or spectral filter, where the scaled or normalized amplitude of the light response (i.e. photoreceptor sensitivity) is plotted as a function of the wavelength of the stimulating light.

When there are at least two (or more) types of photoreceptors with photopigments that have different spectral sensitivities in the same animal or eye, the existence of colour vision is possible. A large number of original papers and reviews are available on spectral sensitivities and colour vision in insects and other invertebrates (e.g. Menzel 1979, Hardie 1986, Menzel & Backhaus 1989, Briscoe & Chittka 2001, Kelber 2006) and there is no need to review them here any further. However, what little is known about the spectral sensitivity of the cockroach visual system is given a more detailed presentation later (section 2.3.3).

To summarize this section, it can be said that generally the acuity and spatial resolution (or spatial filter) is set by the optical properties of the ommatidia and photoreceptors. It can be described through the receptive field and the acceptance angle Δρ. On the other hand, the spectral filter of the photoreceptors is mainly defined by the visual pigments and their spectral sensitivities.

2.1.2 Phototransduction and graded light responses

Each photon absorbed by a photopigment molecule triggers the transduction cascade, a series of biochemical reactions on the rhabdomeric membrane, and produces a single membrane voltage response called quantum bump (Fuortes & Yeandle 1964, Scholes 1964, Dodge et al. 1968, Lillywhite 1977, Lillywhite & Laughlin 1979, Laughlin 1981, Laughlin & Lillywhite 1982, Laughlin 1990,

21

Henderson et al. 2000, Burton 2006). Rushton’s principle of univariance (Naka & Rushton 1966, Rushton 1972) states that after being absorbed the photons have equal values, i.e. different colour photons produce similar quantum bumps. The biochemical events in the transduction cascade of Drosophila are known almost completely (Hardie 2001, Hardie & Raghu 2001, Hardie 2006, Minke & Parnas 2006). Only the last steps before opening the light-gated ion channels are still not clear in the smallest details. There is some evidence that the transduction cascades of other insects and invertebrates might be somewhat different (reviewed by Hardie 2006), but our knowledge about this is not as detailed. The final steps in the biochemical reactions of the transduction cascade lead to opening of light-gated ion channels in the rhabdomeric membrane. This allows ionic current to pass through and causes a voltage change over the membrane, producing the quantum bump. It is generally hypothesized that each individual quantum bump is produced by an individual rhabdomeric microvillus (Howard et al. 1987).

When a photoreceptor is illuminated, the quantum bumps generated by different microvilli propagate along the photoreceptor membrane and sum up to form a graded receptor potential over the photoreceptor soma membrane (Dodge et al. 1968, Wong & Knight 1980, Wong et al. 1980, Wong et al. 1982, Juusola et al. 1994, Juusola & Hardie 2001a, Faivre & Juusola 2008). The size, shape and temporal properties of the bumps are dependent on the average number of incident photons per unit time, i.e. intensity of the light. This means that the phototransduction cascade is modified by light adaptation. Hence, the summation process and graded responses are changed with increasing light adaptation.

One way to study and describe the summation of quantum bumps is to record so-called V-logI functions, where the maximum amplitude of the graded response (Vmax) to some transient stimulus is plotted as a function of the logarithm of the stimulus intensity (I) (Laughlin & Hardie 1978, Laughlin 1981, Matic & Laughlin 1981). As a whole these curves behave nonlinearly. One reason for this is the “self-shunting” of the current through the light-gated channels. The depolarising membrane voltage (Vm) of the cell membrane approaches the reversal potential of the light current (Erev) and reduces the electromotor force (Vm – Erev) driving the ionic current. The membrane current (Im) does not grow relatively as much as the voltage, since

( )m m m revI g V E= − ,

22

where gm = membrane conductance. The fact that a relatively smaller portion of photons is absorbed with an increasing amount of light (i.e. there is a lowering of quantum efficiency) is one additional source of nonlinearity in the formation of receptor potential (Smola 1976, Wu & Pak 1978, Laughlin 1981, Howard et al. 1987, Pelli 1990). The change in the quantum bumps themselves, whereby they get smaller and faster with increasing light adaptation (in addition to the self-shunting effect), is another source of nonlinearity (e.g. Dodge et al. 1968, Smola 1976, Wu & Pak 1978, Juusola & Hardie 2001a, Faivre & Juusola 2008). The presence of outwardly rectifying voltage-gated channels causing changes in gm while Vm is altered (Weckström et al. 1991, Weckström et al. 1995, Weckström & Laughlin 1995) is yet another source.

When only “self-shunting” is taken into account, the V-logI curve can be fitted with the Lipetz (1971) equation. But when all the additional sources of nonlinearity are also considered, the additional exponent n is added to the equation (with self-shunting n is equal to one) (Lipetz 1971, Laughlin 1981). This gives a modified version of the Lipetz equation:

max( )

( ) 1

n

n

RIV VRI

=+

,

where R = reciprocal of the light intensity (I) producing a half-maximal (Vmax/2) amplitude for the light response. The exponent n is dimensionless and has values smaller than one. The smaller the value of n, when fitted to experimental data, the more nonlinear is the summation of quantum bumps and formation of receptor potential.

2.1.3 Adaptation

Generally adaptation at the photoreceptor level means the optimization of sensitivity according to the intensity level of the ambient illumination (reviewed e.g. by Autrum 1981, Laughlin 1981, Laughlin 1989). There are two basic mechanisms of adaptation in the photoreceptors. The photochemical adaptation modifies phototransduction and the electrical properties of the photoreceptor membrane and is the faster mechanism taking place usually within seconds. The slower photomechanical adaptation can last several minutes and causes morphological changes in the photoreceptors.

23

In the compound eyes of Dipteran flies the photomechanical adaptation takes place through the migration of pigment granules inside the photoreceptors (Kirschfeld & Franceschini 1969). These granules aggregate around the rhabdom in light adaptation and the mechanism is thus called an intracellular pupil. This reduces the amount of light travelling inside the rhabdom and thus also the number of photons absorbed by photopigments (Snyder & Horridge 1972, Snyder 1975, Roebroek & Stavenga 1990b) (cf. section 2.3.3). This again means lowered sensitivity for light, i.e. more light is needed to produce as large responses in light adaptation as was produced by dimmer stimuli in dark adaptation. As a result light adaptation causes the V-logI curve to move towards higher intensities along the logI-axis (Laughlin & Hardie 1978, Laughlin 1981, Matic & Laughlin 1981, Roebroek & Stavenga 1990a).

Photochemical adaptation for its part alters the size, shape and latency of the quantum bumps (e.g. Dodge et al. 1968, Smola 1976, Wu & Pak 1978, Laughlin 1981, Juusola et al. 1994, Juusola & Hardie 2001a, Faivre & Juusola 2008) and also the properties of the photoreceptor membrane (Weckström et al. 1991, Weckström et al. 1992, Juusola & Weckström 1993, Weckström & Laughlin 1995, Krause et al. 2008b). With increasing light adaptation the quantum bumps generally become smaller and faster, and the cell membrane allows faster changes in voltage. This not only makes the visual system faster (see the next section 2.1.4.), but also lowers the sensitivity (as mV:s per photon) via smaller bumps. Again the V-logI curves are shifted towards higher intensities (Laughlin & Hardie 1978, Autrum 1981, Laughlin 1981, Matic & Laughlin 1981, Roebroek & Stavenga 1990a). Their shape can also change, since the value of n in the modified Lipetz function (see section 2.1.2) can be altered.

Effectively, the existence of these adaptation mechanisms means that the functional range of the photoreceptors is always regulated according to the average intensity of the stimulating light. This enables animals to adapt to and cope with the very large variations in the ambient illumination (e.g. Warrant 1999, Warrant 2004, Warrant 2006), even though the range of possible response amplitudes or voltages is very limited (about 60-70 mV) in any one adaptational state.

2.1.4 Temporal resolution

Temporal resolution describes how rapid changes in the intensity (contrast coding) or in any other property of the light can be coded into the changes of the light

24

response signals. It can be evaluated either through recording the impulse responses of the photoreceptor (Howard et al. 1984, Laughlin & Weckström 1993, Laughlin 1996, see also 2.3.3) or by a more common approach via linear systems or white noise analysis in the frequency domain (e.g. French 1980c, Juusola et al. 1994, Burton et al. 2001, Juusola & Hardie 2001a, Niven et al. 2003, Vähäsöyrinki et al. 2006, Faivre & Juusola 2008, Frederiksen et al. 2008). The temporal resolution of a photoreceptor is affected and set by the temporal properties of both phototransduction and the cell membrane (see also Weckström et al. 1992, Juusola & Weckström 1993, Laughlin & Weckström 1993, Weckström & Laughlin 1995).

In addition, as discussed above (2.1.3), temporal properties of light responses can also be influenced and regulated by the level of light adaptation. The modification of the temporal properties of phototransduction and quantum bumps on the basis of light adaptation has been shown and described using the white noise approach in many species: e.g. Calliphora (Juusola et al. 1994, Juusola et al. 1995), Drosophila (Juusola & Hardie 2001a, Juusola & Hardie 2001b, Niven et al. 2003), other Dipteran flies (French 1979, Burton 2006), locusts (Kuster & French 1985, Faivre & Juusola 2008), the bee Megalopta (Frederiksen et al. 2008) etc. In all these studies it has been shown that phototransduction speeds up with increasing illumination, meaning faster bumps with shorter latencies and faster contrast coding. It is also evident on the basis of these studies that, although the amplitude of bumps gets smaller with increasing light adaptation, the contrast gain increases. This all means that the signal-to-noise ratio (SNR) of the photoreceptor light responses improves with increasing levels of ambient light, and so does the information capacity (information carried by the responses in bits/s) as well. This even occurs in the nocturnal bee Megalopta, though it normally prefers being active in dim or dark conditions.

In addition to the above described changes in phototransduction, all the voltage responses, quantum bumps and graded responses are further modified or filtered by the properties of the cell membrane (Weckström et al. 1991, Weckström et al. 1992, Laughlin & Weckström 1993, Weckström & Laughlin 1995, Niven et al. 2003). In the first place, the passive properties of the membrane, the resistance (Rm) and the capacitance (Cm), define the time constant (τm) of the membrane:

m m mR Cτ = .

25

Secondly, when the membrane voltage (Vm) changes, the voltage-gated ion channels (most often K+ channels) in the soma membrane further modify the responses by altering the resistance, or conductance (gm = 1/Rm), of the membrane. Basically, the increased opening of these channels with the depolarisation of the photoreceptor membrane in light adaptation increases gm (or lowers Rm) and shortens τm allowing faster modifications of the membrane potential. The membrane impedance (Rm as a function of frequency) has also been evaluated with the white noise approach at different voltages (Weckström et al. 1992, Juusola & Weckström 1993). It has also been shown by this method that the photoreceptor membrane bandwidth increases, and the upper cut-off frequency rises towards the higher frequencies, with increasing depolarisation. The dependence of gm on Vm can be more exactly modelled with the aid of Hodgkin-Huxley-type equations (Niven et al. 2003, Kauranen & Weckström 2004, Vähäsöyrinki 2004). This modelling can also be utilized in evaluating the evolutionary and adaptive effects, and also of the different channels types, on the coding properties of the photoreceptors, and also in other neurons. Recently it has been shown that the properties of the voltage-gated K+ channels responsible for regulating the membrane impedance can also be directly regulated or affected by the phototransduction cascade (Krause et al. 2008b). The photoreceptor voltage signals are even further modified by cable effects while propagating along the photoreceptor membrane (Johnston & Wu 1995, Koch 1999). When different types of voltage-gated channels exist in the output part of the photoreceptors, like in the axons of honeybee drone (Coles & Schneider-Picard 1989, Vallet et al. 1992, Vallet & Coles 1993), the temporal properties of the signal conduction can be modified and further enhanced.

To summarize this section, the temporal resolution of photoreceptors is defined first by the phototransduction cascade producing the photocurrents through the light-gated ion channels. Secondly, it is modified by the current-to-voltage conversion, which is shaped by the membrane impedance, i.e. both by the properties of the passive membrane and the voltage-gated ion channels. In addition, light adaptation plays a key role in adjusting not only the temporal properties of the phototransduction, but also the membrane filter.

It is also useful to note that photoreceptors code all these above described changes to their membrane potential (sections 2.1.1-2.1.3) without relaying any information in the voltage signal about the properties of light that are actually changing (intensity, direction, colour etc.). To determine whether the change was spatial, spectral or temporal is a task for the whole neural network of vision and

26

the brain. This can be done only by comparing the responses from different photoreceptors. There is also ample information available about the processing of visual information beyond the photoreceptors, i.e. in the optic ganglia of the compound eyes (e.g. (Shaw 1984, Laughlin 1987, Stavenga & Hardie 1989, Juusola et al. 1995, Uusitalo 1995, Egelhaaf & Warzecha 1999, Egelhaaf et al. 2002, Egelhaaf 2006, Strausfeld et al. 2006). They are not reviewed here any further, except in the case of what is known about this in the cockroach (presented later in section 2.3.3).

2.2 Vision in dim light

Extensive reviews on dim light vision, covering both vertebrate and invertebrate solutions for seeing in low illumination, do exist (Land 1981, Hess et al. 1990, Laughlin 1990, Warrant 1999, Warrant 2004, Warrant 2006, Frederiksen 2008, Warrant 2008) and there is no need to repeat them here. Only some major points and problems for dark vision, concerning especially the case of the cockroach, are reviewed here. The known specialisations for dim light vision in the cockroach are given a more detailed treatment later (sections 2.3.2 and 2.3.3).

2.2.1 Noise

Seeing or reliable coding and processing of visual information in a dim or dark environment is a self-contradictory task, since the information carried by dim light is unreliable by nature (reviewed e.g. by Warrant 2006, Frederiksen 2008). This is because in low light conditions the light signals themselves are severely contaminated by photon shot noise. The quantum nature of light (photons) leads to random (Poisson-distributed) arrival of photons at any given point of space. When the number of incident photons (N) in a given time interval and light intensity (I) is low, the variance (due to randomness) of incident photons (√N or √I) is large in relation to the number of incident photons itself. Thus the dimmer the environment the smaller is the SNR of the light signals:

N ISNR N IN I

= = = = .

From the viewpoint of photoreceptors this is referred to as extrinsic noise that is due to the light source. A second unavoidable source of noise in visual processing is the noise generated by the phototransduction cascade. This in turn is referred to

27

as the intrinsic or transducer noise (Lillywhite & Laughlin 1979, Laughlin & Lillywhite 1982), which, by its nature, is also Poisson-distributed random shot noise. Variations in the biochemical events of the transduction cascade produce transducer noise by causing the quantum bumps, even in a given state of adaptation, to have different latencies, amplitudes and durations (see also Wong et al. 1980, Laughlin 1990, Henderson et al. 2000, Juusola & Hardie 2001a, Faivre & Juusola 2008). This means that the membrane signals of any photoreceptor in dark or extremely dim conditions are unavoidably noisy.

To be able to see in dim light one has to somehow overcome the noisiness of the light responses. There are two basic ways to do this (reviewed by Laughlin 1990, Warrant & McIntyre 1992, Warrant 1999, Warrant 2004, Warrant 2006, Frederiksen 2008). The first is temporal summation via integrating or summing the incident photons for a longer time. The second is spatial summation, by collecting light from a wider area in space.

2.2.2 Temporal and spatial summation

Basically, the subject of temporal summation at the photoreceptor level has already been covered earlier (section 2.1.4). To recapitulate, it can be said that while temporal resolution is improved with light adaptation, the reverse is also true, i.e. temporal summation increases with dark adaptation via slower phototransduction (e.g. Kuster & French 1985, Juusola et al. 1994, Juusola & Hardie 2001a, Faivre & Juusola 2008, Frederiksen et al. 2008) and a slower membrane (e.g. Weckström et al. 1992, Juusola & Weckström 1993, Laughlin & Weckström 1993, Weckström & Laughlin 1995). Nocturnal, crepuscular and dark-active animals usually have slow impulse responses (Howard et al. 1984, Laughlin & Weckström 1993, Laughlin 1996) and slow and large quantum bumps (reviews e.g. Laughlin 1981, Laughlin 1990) when dark adaptated. This means both long integration times of light responses and a high gain of phototransduction. However, many diurnal or crepuscular species seem to have almost as slow light responses when dark-adapted as the more nocturnal species (Howard et al. 1984, Laughlin & Weckström 1993).

Spatial summation strategies include two basically different approaches (Warrant 1999, Warrant 2004, Warrant et al. 2004, Warrant 2006). The first is the possession of large receptive photoreceptor fields via various means, and the second is neural summation or pooling from several photoreceptors. Most nocturnal insects use the former and have optical superposition eyes (Laughlin

28

1990, Warrant 1999, Warrant 2004, Warrant 2006). This is because the superposition compound eyes have specialized optical and structural features to gather or spatially pool light efficiently. For example the moth Deilephila elpenor is known to be able to fly and even see colors in star light intensities (Kelber et al. 2002). The clear zones of superposition eyes below the lenses and in front of the photoreceptors, assisted by the migrating pigments, make it possible to collect light through several lenses for each of the photoreceptors. In dark adaptation the screening pigments between the ommatidia are removed enabling massive spatial summation of light. In light adaptation in some species the ommatidia can again be optically isolated, when the pigments migrate back between them, and good spatial resolution is enabled.

On the other hand, apposition eyes, like those of the cockroach (Trujillo-Cenoz & Melamed 1971, Wolken 1971, Butler 1973b), are generally considered to be badly suited for vision at low luminance (Laughlin 1990, Warrant et al. 2004, Warrant 2006, Warrant 2008). Large dark-adapted receptive fields (e.g. Butler & Horridge 1973a, Warrant et al. 2004) and other structural specialisations (e.g. Butler 1973a, Butler 1973b, Greiner et al. 2004a, Greiner 2006, Warrant 2006) for collecting light effectively for each of the ommatidia, improve photoreceptor sensitivity. On the other hand, these specialisations are probably not even nearly enough achieve proper night or dim light vision. Yet, in addition to cockroaches, there also seem to be other exceptions. Some apposition-eyed bees, wasps and butterflies are also known to have a nocturnal way of life (Warrant et al. 2004, Warrant 2008). Probably the only way left for these animals to have reliable vision in low light conditions is neural pooling, summation of light information from several photoreceptors (see also Warrant 1999, Warrant 2006). On the basis of anatomical findings (e.g. Ribi 1977, Greiner et al. 2004b, Greiner et al. 2005)) this is thought to happen in the first synaptic area of the visual tract, the lamina, but the electrophysiological evidence for this is still lacking.

2.3 Vision of the cockroach

Among winged insects (Insecta; Pterygota) cockroaches form one of the phylogenetically oldest groups. The oldest fossil discoveries are close to 400 million years old (Kambhampati 1995). The suborder Blattodea (or Blattaria; cockroaches) forms an order (Dictyoptera) together with the Mantodea (praying mantises) and includes several lower taxonomic groups and all together over 4000 species (Bell 1990, Kambhampati 1995, Grandcolas 1996). The particular species

29

studied in this thesis, i.e. the American cockroach (Periplaneta americana), belongs phylogenetically, according to the fossil records, morphology, DNA mapping and breeding strategy, to one of the oldest (Blattoidea; Blattidae; Blattinae) of the taxonomic groups among the Dictyoptera.

Cockroaches are widely studied and there are several books written on cockroach biology, ecology, evolution, physiology and neurobiology (e.g. Guthrie & Tindall 1968, Bell & Adiyodi 1982, Huber et al. 1990a, Huber et al. 1990b, Bell et al. 2007). Particularly the American cockroach is a widely used experimental animal, but not really very much is known about its vision, especially in the sense of what the animal needs or uses it for. On the basis of studies on its behaviour and sensory systems, it can be readily concluded that the other sensory systems, not vision, are primary sources of information for cockroaches. The main sensory systems controlling their behaviour and locomotor activity, especially the widely studied escape behaviour, are the different mechanosensory organs and cells in their cerci (e.g. Camhi 1980, Camhi & Levy 1989, Levi & Camhi 2000), antennae (e.g. Burdohan & Comer 1996, Ye & Comer 1996) and legs (e.g. Zill 1990, Zill & Seyfarth 1996), and additionally the chemosensory system also mainly located in the antennae (e.g. Seelinger 1990, Willis & Avondet 2005, Willis et al. 2008). Probably the most preferred form of locomotion for Periplaneta is walking or running and following a wall or some other similar surface by tapping or sweeping it constantly with the antenna ipsilateral to the surface (Camhi & Johnson 1999, Cowan et al. 2006). The escape responses are primarily evoked by air movements and vibrations sensed by the cerci (Camhi 1980, Camhi & Levy 1989) or unpleasant contacts with the antennae (Comer et al. 2003, Ye et al. 2003). On the other hand, Periplaneta is one of the fastest runners (in relation to its size) in the world (Camhi & Johnson 1999) and they can also fly (Goldstein & Camhi 1988, Libersat & Camhi 1988), though only weakly and only in high enough temperatures. So, even though mechanosensory information probably plays a crucial role in performing these functions, in principle these kinds of behaviours could benefit from seeing and the ability to detect motion.

Consequently, on the basis of what is known, while performing its usual behaviour in the conditions where the cockroach normally prefers to live, the visual system seems to be some kind of secondary or aiding sensory system for the primary or most important mechano- and chemosensory systems. However, a fair number of studies have been published about cockroach vision (reviewed by Mote 1990). Cockroaches have large, structurally and functionally complex,

30

compound eyes (e.g. Butler 1973a, Butler 1973b, Butler & Horridge 1973a, Butler & Horridge 1973b, Heimonen 1990, Mote 1990, Heimonen 1999), and large ocelli (e.g. Mizunami 1994, Mizunami 1995a). It is possible to ask, why do “creatures of darkness” like cockroaches have large eyes and what do they need and use their vision for? How and what, if anything, is it possible to see in the dim and dark conditions they mostly prefer to live in?

2.3.1 Visually guided behaviour of the cockroach

To start with, a further question can be asked: “Is there any visually or light guided behaviour in the cockroach?” A short answer on the basis of what has been known already for a fairly long time would be: “Yes, they regulate their daily locomotor activity rhythm on the basis of visual input via the compound eyes” (e.g. Roberts 1965, Lipton & Sutherland 1970, Ball 1971, Rivault 1983, Mote 1990, Page 1990). They are most active for a few hours after the dark period commences and sometimes (or some individuals) also before or after the light period sets in. Large compound eyes, looking simultaneously almost everywhere in the surroundings of the cockroach (Butler 1973a), would not be fundamental for the regulation of circadian rhythms. Smaller eyes, or ocelli, would be quite enough. Since the large eyes exist, it seems that it is a task for the green sensitive photoreceptors in the compound eyes to provide information for the circadian behavioural rhythms (Mote 1990).

The next question would be: “Are cockroaches afraid of light?” It has long been known that several cockroach species show negative phototaxis, i.e. they prefer dim and dark environments over brightly illuminated ones (e.g. Guthrie & Tindall 1968). It has also been shown that cockroaches avoid illuminated spots when walking on an arena (Kelly & Mote 1990a). This avoidance is colour-dependent, and on bright backgrounds cockroaches maximally avoided (ca. 80% of trials) short wavelength light (400 nm). In contrast, on dim backgrounds longer wavelength lights (540-570 nm) were most strongly avoided, though only ca. 40% of trials provoked avoidance. In addition, it has been shown that visual input, such as a flash of light, in conjunction with an air puff directed to the cerci, can modify preferred escape directions of the cockroach (Riemay 1984). But the escape itself is evoked by air currents or puffs, not by the light alone. Recently, while studying social aggregation behaviour and the changes robots can induce to this behaviour (Halloy et al. 2007), it was also shown that Periplaneta is not afraid of light and can even learn to prefer more luminous places over the dimmer

31

ones. Although innately most cockroaches aggregated below dim shelters in the course of time (not immediately, put in tens of minutes), they could be induced to change this collective behaviour. As soon as pheromone flavoured robots among them were programmed to choose more brightly illuminated shelters over the dim ones, the cockroaches did so too. So, cockroaches are not afraid of light as such, but they do normally prefer dim or dark places, if given a chance to choose.

The third basic question could be: “Are cockroaches ever actively interested to follow or act on any visual or light signals?” Already over 50 years ago it was reported (Autrum & Stöcker 1952, Autrum 1958), while studying mostly flies, that Periplaneta has an optomotor turning response to spatial frequencies up to about 10 Hz. The existence of this optomotor response has later been confirmed (Kettunen 1994, Voutilainen 1996), even though the results were slightly different and not that clear. (Unfortunately neither of these studies has been published on any international forum.) Kettunen (1994) reported that in about 60-80% of trials tethered and walking male cockroaches had a positive optomotor response immediately after the stimulus was started. Her results were probably seriously contaminated by mechanosensory cues, i.e. vibrations of experimental setup that primarily governed the behaviour. With a more sophisticated setup Voutilainen (1996) showed that in about 90% of trials cockroaches (again walking and tethered) have a positive optomotor turning response, when stimulated with spatial frequencies between 0.5 and 5 Hz. The response was weaker (70-80%) at 10 Hz, and with frequencies of 0.1 and 15 Hz there were no significant differences with the motionless controls. The positively responsive frequency band (0.5-10 Hz) was the same at all intensities of background light tested (ranging 3 log units) except the dimmest, where the highest frequency with positive optomotor responses was 5 Hz (frequencies between 5 and 10 Hz were not tested). The existence of the optomotor response implies an ability to see at least slow frequency motion. On the other hand, flies are very well known to be able to see fast motion (e.g. Egelhaaf & Kern 2002), but still choose to have an optomotor response only to slow spatial frequencies (1-10 Hz) while walking (Pick & Buchner 1979). Maybe this could be true also for the cockroach?

In one study (Mizunami et al. 1998) it was reported that Periplaneta can localize itself in a scene and learn to find a place according to visual landmarks. In this study cockroaches were first placed on a hot (44-47 °C) black arena with a cool (17-20 °C) white place. After a few trials they learned to find this cool white place much faster than originally, or in the case when the cool place was black. In the second part of the study the walls of the arena were decorated with visual

32

landmarks and the arena itself, including the cool place, was completely black. Also in this case most of the cockroaches (ca. 80%) were able to learn to find the cool place faster after a few trials. Moreover, the individuals that learned this were lured to a wrong place, when the landmarks were turned to a different position.

Until quite recently not very much was known or published on any other forms of visually guided behaviour in the cockroach except these above mentioned studies. However, during the last five years a few more visually guided, or at least aided, forms of behaviour have been described. Maybe most importantly some evidence is starting to accumulate on how vision in Periplaneta is used to guide and aid the antennae equipped with the mechano- and chemosensory abilities to find their targets. Studies (Ye et al. 2003, Kwon et al. 2004, Lent & Kwon 2004) show that cockroaches can see and localize objects of interest, if motivated to do so, and that they can even learn to associate visual stimuli with other sensory cues.

At first the influence of visual inputs on antennal reorientations and the directions of escape evoked by mechanically touching the antennae were studied (Ye et al. 2003). On the basis of only monitoring freely behaving and interacting cockroaches (and later tethered ones) it was deduced that vision is used to target the antennae towards moving targets of interest like other approaching cockroaches or presented artificial targets. In additional tests with tethered cockroaches, the escape direction (evoked by an unpleasant touch of an antenna) was affected by the angle form which the antenna was touched. Therefore, since the antenna is directed by using visual cues, vision has an indirect effect on the escape direction. In addition, it was found that covering the compound eyes did not alter the directionality of the escape directly, but it did shorten the average length of the escape run and reduced the typical speed of running. The shade-induced pause or termination in escape running has already been reported earlier (Okada & Toh 1998), and was observed to be mediated via the compound eyes and to take place even in very low background light intensities.

Secondly, an association between visual and chemosensory cues has also been reported (Kwon et al. 2004, Lent & Kwon 2004). Normally, cockroaches actively reach something with their antennae only on the basis of attractive smells (food, pheromones etc.). But they can also be taught to associate this smell with a light signal. After this association is formed they move the antennae toward a light stimulus only. However, it has been noted (and stressed) that lights alone

33

were almost never interesting for untrained animals - they became attractive only after an association with the smell of food was established.

Most of these studies of visually guided behaviour in cockroaches have been done in bright light, because of the demands of video recording. Only a few of them (Kelly & Mote 1990a, Kettunen 1994, Voutilainen 1996, Okada & Toh 1998) even distantly, address questions about the ability or performance of cockroach vision in dim light, but they show that cockroaches use and need their vision for some behavioural tasks. It also has to be noted that the variation in the behaviour of individual American cockroaches has, on several occasions (e.g. Lipton & Sutherland 1970, Kelly & Mote 1990a, Kettunen 1994, Voutilainen 1996), been reported to be very large. This makes them a very demanding target for behavioural experiments.

2.3.2 Structure of the cockroach visual system

The compound eyes of the cockroach Periplaneta are of the apposition type (Wolken 1971, Butler 1973b), phylogenetically and morphologically the oldest type of compound eyes. The other basic types of compound eyes, the superposition and neural superposition eyes, evolved when cockroaches had already existed for a time (Nilsson 1989, Shaw 1989, Shaw 1990, Nilsson & Kelber 2007).The ommatidia of the cockroach eye, like in all apposition eyes, are optically isolated and completely surrounded by the screening pigment cells, which uncommonly also reach between the lenses, i.e. the facets (Wolken 1971, Butler 1973b). This is why, in this special case, the amount of light reflecting out of the eyes is extremely low and the compound eyes are uniformly black.

Cockroaches are hemimetabolous and in an eye of an individual Periplaneta the number of ommatidia increases during growth from about 50-100 to about 2000-3500 (Nowel 1981, Stark & Mote 1981). This increase is almost continuous (not only during moulting) and takes place at the ventral, frontal and dorsal rims of the eyes (Nowel 1981). Also the size of all the ommatidia increases during this development. The number of ommatidia in the eye of an adult cockroach varies individually and correlates with the size of the individual (Füller et al. 1989). Consequently the reported number of ommatidia also varies greatly between studies: 2000 in females and 3000 in males (Walther 1958b); an average of about 2000 (Wolken 1971); over 2000 (Stark & Mote 1981); an average of about 3500 (Nowel 1981). According to the most thorough study (Füller et al. 1989), there is on average 4258 ommatidia per eye in males and the variation between individual

34

eyes is large (3313-5707). The two eyes of a cockroach may also have significant differences in the number of ommatia, e.g. 4495 in the right and 4013 in the left eye.

There are also other kinds of unusual variations in the structure of the Periplaneta compound eye. Usually the lenses of insect compound eyes are organised in regular rows of regular hexagons (e.g. French et al. 1977, Stavenga 1979, Land 1981, Laughlin 1981) and this is thought to be imperative for the proper function of the eye. In Periplaneta some roughly regular rows of nearly regular hexagonal lenses can be found, but in many places these rows end at irregular clusters of irregular polygons of different sizes and shapes (Butler 1973a). Periplaneta is not the only cockroach where this kind of uncommon variation in lenses is reported. Gromphadorhina portentosa, another large dark active cockroach, is also reported to have this irregular variation in the lenses of its compound eyes (Mishra & Meyer-Rochow 2008). In addition to the variation in sizes and shapes of the lenses, the angles between the optical axes of the ommatidia vary in Periplaneta to a surprising extent (Butler 1973a). Both vertical and horizontal angles vary about 10-fold (1-10 degrees). The largest angles are located in the peripheries of the ventral region of the eye and the smallest in the central part of the dorsal region. In comparison, in the blowfly the corresponding variation is between 1,5 and 4,6 degrees (Burkhardt et al. 1966, Järvilehto 1985).

On the basis of the extent, shape and angles between optical axes the total visual field of the compound eyes of Periplaneta is very wide-ranging (Butler 1973a). At its widest it is 240° in the horizontal and 198° in the vertical plane per eye. Hence, the visual fields of the eyes significantly overlap: dorsally by 40°, anteriorly by 65° and posteriorly by 56°. This enables the animal to see almost its whole surroundings simultaneously and possibly even to have binocular vision in the areas of overlap.

The optic lobe of the cockroach resembles that of other insects and includes a retina and three optic ganglia: the lamina, medulla and lobula (Ribi 1977, Mote 1990). The retina is formed by the screening pigment cells and photoreceptor somata, which mainly send their axons into the 1st synapse in lamina. Some of the photoreceptors and all the monopolar neurons of the lamina terminate in the medulla. The structure of the retina is fairly well known (Butler 1971, Trujillo-Cenoz & Melamed 1971, Wolken 1971, Butler 1973b, Ferrell & Reitcheck 1993), but already the retina-lamina connections by photoreceptor axons, and the structure of lamina, are much less characterized (Ribi 1977, Ernst & Füller 1987) and leave many questions open. Concerning the structures and connections from

35

lamina onwards only some scattered information is available (Kelly & Mote 1990b, Mote 1990).

Retina

The retinal ommatidium of the cockroach Periplaneta is composed of a corneal lens, a crystalline cone made of four cells of Semper, and eight retinula or photoreceptor cells, which each have a light sensitive rhabdomere i.e. the microvilli (Trujillo-Cenoz & Melamed 1971, Butler 1973b). The depth of the lenses and crystalline cones is about 150 μm and the retinula cell layer is 250-350 μm thick. The diameter of the ommatidium is 30-35 μm distally, but narrows down to 5-10 μm proximally, close to the basement membrane. The rhabdom, comprised of eight rhabdomeres, is of the closed type, i.e. the rhabdomeres form a joint light guide with a common optical axis (Snyder & Horridge 1972, Snyder et al. 1973). The overall anatomy of ommatidia is similar all over the retina, but the sizes of the photoreceptor cells vary between different locations in the retina (Wolken 1971). The diameter, and thus also the cross- sectional area, of the rhabdom below the crystalline cone also varies between individual ommatidia (Ferrell & Reitcheck 1993). When compared with the day-active blowfly Calliphora the area of the cross-section of the rhabdom is on average about five times larger in the night-active cockroach Periplaneta (Wolken 1971). The cross-sectional area of a rhabdom stays quite constant while advancing distally to proximally, and only just before the basement membrane does it become smaller and disappear (Trujillo-Cenós and Melamed 1971, Butler 1971). In contrary, the cross-sectional area of most individual rhabdomeres changes while advancing along the rhabdom (Butler 1971). In every ommatidium it is distally large in three of the photoreceptors and becomes smaller proximally, but this is the opposite in the other three cells. Every ommatidium has one photoreceptor, where the diameter of the rhabdomere is comparatively small for the whole length of the rhabdom and one where it’s relatively large. According to Butler (1971) it is also possible to number and identify the individual photoreceptors in an ommatidium according to the order with which their rhabdomeres appear in the cross-sectional serial sections. Apart from having a fairly large diameter and length, the cockroach rhabdom has one more special structural feature, namely, it also has a cup or goblet-like structure at its distal end that surrounds the proximal crystalline cone (Trujillo-Cenoz & Melamed 1971, Butler 1973b). This structure on its own

36

ensures that the light captured by the optics is also transmitted to the rhabdom and thus enhances the light-collecting power of the ommatidium.

As first described in locust (Horridge & Barnard 1965), cockroaches also possess a vacuolar structure called the palisade, which is formed around the rhabdom when photoreceptors are dark adapted (Butler 1971, Butler 1973b, Ferrell & Reitcheck 1993). During light adaptation the vacuoles move further away from the rhabdom, which is instead enveloped by migrating pigment granules inside the photoreceptors. The functional correlates of this structural change are described in the next section (2.3.3).

Retina-lamina projection

Deviating from what is generally found in compound eyes, Periplaneta has below the retina two basement membranes (or basement-membrane like structures), which are located 40-130 μm apart from each other in the central parts and connected to each other at the rims of the eyes (Trujillo-Cenoz & Melamed 1971, Ribi 1977, Ernst & Füller 1987). When the rhabdomeres in the photoreceptor somata disappear little before the 1st basement membrane, the photoreceptors start to resemble axons (Trujillo-Cenoz & Melamed 1971, Butler 1973b). Between the basement membranes the eight axons leaving the retina from one ommatidium form a bundle enveloped by a glial sheath (Ribi 1977).

Also contrary to most insects, the distance between the retina and lamina is long (Butler 1973b, Ribi 1977, Ernst & Füller 1987, Mote 1990). The direct distance between the proximal retina and distal lamina is at its shortest 200-250 μm (Ribi 1977, Ernst & Füller 1987) and the photoreceptor axons on average are even longer: ca. 1 mm (Butler 1973b) - 750 μm in the dorsal area and up to 1500 μm in the ventral area of the eye (Ernst & Füller 1987).

According to Ribi (1977) there are two basic types of photoreceptor axons. The more common type synapses in the lamina and some even longer axons terminate directly in the medulla. The axons from each ommatidium penetrate first the basement membranes and the so called fenestrated layer (dense tracheal network) in bundles of eight. Then they reorganize into bundles of 6-20 axons before entering the lamina. The longer axons travel directly through the lamina. The bundles of eight axons below the retina always have one thin axon (diameter 0.8-1.5 μm) and seven thicker ones (2.0-4.4 μm). The axons terminating in the lamina have a diameter between 1.5 and 2.0 μm and the axons passing through the lamina are always thinner (0,8-1,5 μm). Axons leaving the retina in about the

37

same area also usually terminate in approximately the same area in the lamina, but there are also axons deviating from this rule. There are no regularly repeating structures (like the lamina cartridges in the flies) in the lamina of the cockroach. The somata of the monopolar cells are situated distally to the lamina. Monopolar cells are postsynaptic to the photoreceptors and have broad dendritic arborisations in the lamina. This is also typical for many other nocturnal insects and is thought to indicate pooling or summation of many photoreceptor signals onto the same second order neuron (Greiner et al. 2004b, Greiner et al. 2005, Greiner 2006, Warrant 2006, Warrant 2008).

Ernst and Füller (1987) obtained similar results to Ribi (1977), but presented some more detailed results and also some differing discoveries. They found that all the axons have constrictions (diameter only 0.9-1.9 μm), when they penetrate the basement membranes. The diameter of the axons was found to be 2.1-5.5 μm between the basement membranes and 1.5-5.0 μm proximally to them. They also describe slightly different reorganisation of axons into bundles before the lamina, but don’t mention the numbers of axons in these bundles. It can also be noted that the organization of the axons is such that the degree of retinotopy in their projection to the lamina cannot be learned without marking individual axons. The only regular structures in the lamina are the longer photoreceptor axons that pass through the lamina with constant distances to each other.

Unusual irregularities in the structure

When summing up the structure of the visual system of the cockroach, two things are highlighted. First of all, it is obvious that the structure shows several adaptations to a nocturnal or dark-active life style (Trujillo-Cenoz & Melamed 1971, Wolken 1971, Butler 1973a, Butler 1973b, Ribi 1977, Mote 1990, Ferrell & Reitcheck 1993). Secondly, it as evident that the arrangement of the compound eyes includes numerous irregularities and oddly random variations (Butler 1971, Butler 1973a, Butler 1973b, Ribi 1977, Ernst & Füller 1987, Füller et al. 1989, Mote 1990, Ferrell & Reitcheck 1993), which can be described as “structural noise”. It is not only Periplaneta that has these irregularities, but another cockroach is also reported to have them (Mishra & Meyer-Rochow 2008). Hence, this might be a more common feature among the cockroaches, at least the nocturnal ones. However, irregularity is highly unusual for visual systems, where regularity or high order systematic organisation is thought to be of fundamental

38

importance to the performance of vision (e.g. French et al. 1977, Stavenga 1979, Land 1981, Laughlin 1981).

To summarize these irregularities and variations in Periplaneta one can start with the large variation in the number of ommatidia per eye between adult individuals, and even between the two eyes of a single individual (Walther 1958b, Wolken 1971, Nowel 1981, Stark & Mote 1981, Füller et al. 1989). In addition, the shapes and sizes of lenses in an eye vary greatly as do the angles between optical axes (Butler 1973a). The diameter of the rhabdom varies between the ommatidia (Wolken 1971, Ferrell & Reitcheck 1993), as does the length of the rhabdom (Wolken 1971, Butler 1973b) and the size of individual photoreceptors and their rhabdomeres (Butler 1971). Below the retina the irregularities continue. The length and diameter of the photoreceptor axons vary greatly (Ribi 1977, Ernst & Füller 1987). The size, shape and broadness of pre- and post synaptic structures in the lamina differ from each other between cells (Ribi 1977). The number of axons in a bundle entering the lamina varies (Ribi 1977, Ernst & Füller 1987). Numbers of certain cells in the optic lobes, when compared between eyes, are different (Füller et al. 1989), and so on. It is difficult to see the reason for these irregularities and seemingly uncontrolled variations, and it seems obvious to ask why they exist. What can actually be seen with a visual apparatus like the eye of the cockroach?

2.3.3 Function of the cockroach visual system

The cockroach compound eye is classified as a “slow” compound eye on the basis of the flicker fusion frequency, measured from extracellular responses to repeated light flashes (Autrum 1958). In this same study several other insect species were also classified on the same grounds. It was concluded, though not with this terminology, that diurnal and crepuscular animals active in dim or dark environments have “slow” vision, i.e. slow light responses, and integrate light stimuli over long time periods. In contrast, diurnal insects active in bright light, especially flying insects, were found to have “fast” vision.

Temporal integration

In another comparative study with several insect species, based on recordings of dark and light adapted impulse responses and their time-to-peak values, it was shown that when dark adapted all species are more or less as slow (Howard et al.

39

1984), with long integration times. Only houseflies (Musca domestica) were found to be faster than locusts (Locusta migratoria). Astonishingly, for instance, cockroaches (Periplaneta americana) and houseflies had on average the same time-to-peaks for light responses when dark adapted. When light adapted, the housefly was significantly faster than all the other tested species. Moreover, other fast flying species (dragonflies and drone flies) were found to be significantly faster than species having a slower style of life (locusts and crickets). Unfortunately the cockroaches were reported to be so unstable and difficult to record from intracellularly that the light adapted responses were not published. It thus seemed that all insects, even the fast flying diurnal species, have long integration times or slow responses when dark adapted, but that it was only the fast and agile animals that had significantly faster responses when light adapted.

The speeding up of phototransduction with increasing light adaptation has since been confirmed with more elegant methods in several species. It occurs even in the locust (Kuster & French 1985, Faivre & Juusola 2008) and in the nocturnal bee Megalopta (Frederiksen et al. 2008), though with a lesser extent than in fast flying species (Juusola et al. 1994, Juusola & Hardie 2001a). This shows that the integration time of light responses in all these species is regulated and adapts on the basis of ambient illumination conditions. This regulation enables them all to see something in a dim environment. On the other hand, it also increases the information capacity, i.e. the amount of information carried per unit of time (in bits/s), of the photoreceptors in the bright conditions. The faster the transduction (i.e. the wider the band width) the more information that can be carried (see e.g. Shannon & Weaver 1949, Juusola & Hardie 2001a), given that the SNR is either constant or increases with the intensity as well.

What about the behaviour of temporal summation with increasing light adaptation in the cockroach? It is known that in dark adapted cockroach photoreceptors the quantum bumps, as well as the impulse responses, are large and slow (Smola 1976, Laughlin 1981, Howard et al. 1984), but very little is known about how the bumps or their summation into receptor potentials behave with light adaptation. It can be seen from the data presented by Butler and Horridge (1973a) that the light-adapted impulse responses are at least to some extent faster than the dark-adapted ones. Smola (1976) reported that the properties of light-induced noise change with increasing light adaptation indicating changes in the quantum bumps.

40

Spectral sensitivity

Already half a century ago it was reported (Walther 1958a) that Periplaneta had a peak in spectral sensitivity at the wavelength of green light (507 nm) and that the sensitivity also increased towards the violet end of the spectrum. It was concluded that in the visual system of the cockroach there are two photoreceptor systems, which differ in their spectral sensitivity. Later this was confirmed and more exactly described (Mote & Goldsmith 1970, Mote & Goldsmith 1971) with intracellular recording and staining in the white eyed mutants of the American cockroach. In these studies it was shown that there are two spectral types of photoreceptors. The UV receptors have a peak in sensitivity at 365 nm and their sensitivity falls very fast at longer wavelengths. The green receptors have a sensitivity peak at 500-510 nm and also some sensitivity in the UV range. Through staining of the photoreceptors by intracellularly injecting them with two dyes of different colours they were able to show that these two spectral receptor types existed within the same ommatidium.

Based on the number of random recording electrode penetrations it has been reported (Mote & Goldsmith 1970, Mote & Goldsmith 1971) that in the dorsal part of the eye there were almost as many UV as green receptors (11:13). Then again, on the same basis it has also been argued this same relation is 4:25 (Heimonen 1990). The location and number of spectrally different photoreceptors has also been studied through microscopy (Butler 1971) by utilising selective light adaptation with either green or UV light and palisade formation in the photoreceptors due to light adaptation. On the basis of identifying the photoreceptors anatomically (as explained in the previous chapter, 2.3.2) it was reported that the relative number of UV to green receptors is 3:5 in each ommatidium throughout the whole retina. Different spectral types always had the same position in relation to each other in every ommatidium and the eyes were mirror images of each other in this respect. In contrast to these findings, Ribi (1977) hypothesized that all the green receptors have a synapse terminal in the lamina and UV receptors in the medulla. Mote (1990) also claims in his review to have shown this in his own (unpublished) studies. They also both strongly imply that there is only one UV receptor, i.e. a photoreceptor with a terminal in the medulla, per ommatidium. On the grounds of these controversies it can be summarized that the relative number of green and UV receptors is still an open question, as well as the functional or spectral identity of the photoreceptors terminating in the medulla.

41

Receptive fields, photoreceptor sensitivity and adaptation

The acceptance angle (Δρ) of dark-adapted green receptors in Periplaneta has been reported to be 6.7°±1.8° (mean ± SD) in the horizontal and 6.9°±1.3° in the vertical plane (Butler & Horridge 1973a). In the light-adapted state the corresponding values of Δρ were markedly narrower: 2.4°±0.9° and 2.3°±0,6°. These changes in Δρ are explained by the clustering of pigment granules close to the rhabdoms in the light-adapted state and the formation of a palisade during dark adaptation (Snyder & Horridge 1972, Butler 1973b). By lowering the optical density of the tissue around the rhabdom, the palisade increases the difference in the refractive indices between the rhabdom and its surroundings. Since the rhabdom acts as a light guide, this leads to incident light being accepted by the rhabdom over a wider angle.

Butler & Horridge (1973b) have also studied the effect of light adaptation on the sensitivity of cockroach green receptors. They recorded V-logI curves and on the basis of the stimulus intensity values inducing half-maximal responses noticed that even in the same state of adaptation the photoreceptors had roughly 10-fold differences in their sensitivity. In addition, on average the dark-adapted cells were about ten times more sensitive than the light-adapted ones. Disappearance of the palisade in light adaptation (Butler 1971, Butler 1973b) explains only about 25% of the drop in sensitivity (Snyder & Horridge 1972). The rest, 75%, is probably explained by the clustering of the pigment granules around the rhabdom. According to theoretical calculations, the pigments in these granules should be about 70 times more efficient in absorbing light than the visual pigments, in order to induce this sensitivity drop.

The green receptors of cockroaches have also been reported to be polarisation sensitive (Butler & Horridge 1973b), i.e. to be about five times more sensitive to light polarized in the optimal plane than to light polarized to the plane orthogonal to this optimum. In addition, photoreceptors could be divided into two groups having optimum planes of polarization 90° apart. These two optimum planes were supposed to correspond with the two orientation planes of the rhabdomeric microvilli. On the other hand, it has also been reported that only some (about 25%) of the photoreceptors are polarisation sensitive and most of them are not (Heimonen 1990).

42

Processing of light information in the optic lobes, CNS and ocelli

Very little is known about information processing in the optic ganglia of the cockroach. There are no reports of any recordings or functional studies on the lamina. And though there are some electrophysiological studies on cockroach medulla (Mote et al. 1981, Mote & Rubin 1981, Kelly & Mote 1990b), the findings are scattered and the general picture is very unclear (review by Mote 1990). Neurons in the medulla respond to light stimuli with action potentials, some with ‘on’-responses and others with ‘off’-responses. Many neurons spiked only when lights were on, others continuously, with only the spiking frequency modulated by the light. On the other hand, neurons in the medulla were colour sensitive and also had largely variable latencies and receptive fields. There are no publications on the function of cockroach lobula.

Large calycal giant neurons terminating in the mushroom body of the cockroach brain have been examined by intracellular recording and staining (Nishino & Mizunami 1998). They exhibit spontaneous and rhythmic burst of spikes, which were suppressed by olfactory, visual, tactile or air current stimulation. The mushroom bodies of the cockroach are also shown to participate in place memory (Mizunami et al. 1998) like that described in more detail earlier (in section 2.3.1). A few studies have appeared on the electrophysiology of neurons responding to light stimuli in the cockroach ventral nerve chord in the thorax (Cooter 1973, Edwards 1982a, Edwards 1982b). These neurons also reacted to light stimuli either by bursts of action potentials or by changes in the frequency of continuous spiking. Cooter (1973) found several different types of neurons responding to light stimuli, some only to ‘lights off’ and some only to ‘lights on’ stimuli, and others only to moving lights. A few of the cells responded only, or also, to ocellar light stimulation and some also to air currents directed toward the antennae. Edwards (1982a, 1982b) in turn did some studies on the so-called descending contralateral movement detectors, which are neurons that react only to movements of the light stimulus. He found that they respond both to UV and green light, with no colour discrimination. Instead, the neurons segregated the intensities of moving lights and adapted on the basis of background light intensity.

The anatomy, and even the electrophysiology, of Periplaneta ocellar neurons are known to a much greater detail (reviewed by Mizunami 1994, Mizunami 1995a) than the corresponding features in the compound eyes. Periplaneta has two white (non- pigmented) ocelli situated frontally on the head between the antennae. There are over 10000 photoreceptors in an adult ocellus (Toh & Yokohari 1988) and on the basis of ERG recordings they are all reported to be green receptors (Goldsmith & Ruck 1958).

43

All the photoreceptors synapse onto only four second-order neurons, which for their part terminate onto 15 third-order cells (Mizunami et al. 1982, Mizunami & Tateda 1986, Mizunami 1994, Mizunami 1995c, Mizunami 1995b). The 2nd and 3rd order neurons have been described in detail not only in structure, but also in function (Mizunami et al. 1986, Mizunami & Tateda 1988b, Mizunami & Tateda 1988a, Mizunami 1990, Mizunami 1994, Mizunami 1996). They have connections to the lamina, medulla and lobula of the compound eyes, to other sensory ganglia in the head, motor ganglia and several areas in the brain. The light responses of the 2nd order cells are membrane hyperpolarisations with off-spikes and those of 3rd order neurons are depolarisations, hyperpolarisations, action potentials and their combinations. Some of the 3rd and 4th order neurons in the ocellar tract are multimodal and react also to stimuli given to other sensory systems, e.g. the mechanosensory system of the antennae (Ohyama & Toh 1986, Mizunami 1995a). On the other hand, although there is ample information on the structure and function of the ocellar neurons, very little known about what the ocelli are used for in the cockroach (Mizunami 1996).

Spatial integration

In addition to slow responses and temporal summation during dark adaptation (Smola 1976, Laughlin 1981, Howard et al. 1984, see also above), the cockroach visual system seems to utilize the strategy of collecting light in space i.e. spatial summation. The first method to spatially integrate low intensity light signals is to have wide photoreceptor receptive fields in dim or dark conditions (Butler & Horridge 1973a). This is accomplished by wide and large ommatidia, and especially by wide rhabdoms (Trujillo-Cenoz & Melamed 1971, Wolken 1971, Butler 1973b), with some additional structural features to insure maximal light capture. On the other hand the receptive fields in dark conditions are even wider because of pigment migration and palisade formation around the rhabdom (Butler 1971, Snyder & Horridge 1972, Butler 1973b, Ferrell & Reitcheck 1993).

The other strategy of spatial integration is the summation of response signals from several photoreceptors to each of the 2nd order neurons. There are several anatomical findings that suggest that a fair quantity of spatial pooling takes place in the 1st synaptic area of the cockroach visual system. According to Ribi (1977) photoreceptor axons regroup on their way to the lamina to form bundles of 6-20 axons. The postsynaptic arborisations of the 2nd order monopolar cells in the lamina are also wide and large, as is typical of nocturnal species (Greiner et al. 2005, Greiner 2006, Warrant 2006, Warrant 2008). Ernst and Füller (1987) also

44

reported that regrouping takes place in the photoreceptor axons before lamina, but they did not give any exact figures on the number of axons in the new bundles. Instead, they measured that the surface area of distal lamina is less than 10% of the area of retina. It has also been reported (Füller et al. 1989) that cell counts in the lamina and medulla are far smaller (about the order of 2000-3000 per ganglion) than the number of photoreceptors per eye (26000-45000). The exact answer to what extent and how this pooling of visual information in cockroach compound eyes occurs is still missing. In the cockroach ocelli it is known that over 10000 photoreceptors are pooled to only four 2nd order neurons (Toh & Yokohari 1988, Mizunami 1995c, Mizunami 1995b). In the medulla of the compound eye the quantity of spatial summation is probably even larger than in the lamina, since the receptive fields of medulla neurons have been reported to be very wide (Mote et al. 1981). All this spatial summation of light information naturally and inevitably leads to a decrease in spatial resolution.

45

3 Aims of the research

The general aim of this thesis has been to reveal the functional features by which the photoreceptors of the American cockroach (Periplaneta americana) have adapted to cope with dim light conditions. The American cockroach was chosen to be the target of the study, since it was abundantly and easily available, there was already some basic information on its visual system and it seems to be an evolutionary success story. There are two things that made Periplaneta americana exceptionally interesting and challenging to study. First of all, it prefers living in dark or extremely dim environments, but has still retained large and complex compound eyes in the course of evolution. Secondly, several peculiarities and mysteries, e.g. structural irregularities, were known to exist in its visual system without any proper explanation.

The more specified aims of the original articles upon which this thesis is based on were:

I To map the functional correlates of the known irregularities in the structure of the cockroach compound eye, i.e. to quantify the possible variability in the properties of light responses and to formulate a theory how this kind of visual system could work.

II To properly describe the temporal summation properties or dynamics of the light responses of cockroach photoreceptors at different levels of light adaptation and to determine if they are matched to low levels of ambient illumination.

III To determine and describe the light response properties in the long axons of the cockroach photoreceptors.

46

4 Summary of the material and methods

4.1 Animals and preparation

In this thesis the green sensitive photoreceptors (Mote & Goldsmith 1970, Mote & Goldsmith 1971) of the adult male cockroach (Periplaneta Americana; Fig. 1A) compound eye have been studied. A few females were also tested to ensure that the light responses and their variability did not differ significantly from males. Cockroaches were maintained in a 24 h (12:12) light-dark cycle either at room temperature (20-22 °C) or at 25-27 °C. Recordings were done during different phases of the circadian rhythm and always at the room temperature.

For capturing, the cockroaches were anaesthetized with CO2. When the effect of the anaesthesia started to fade out and the antennae and the legs started moving again, the cockroach was dissected behind the front legs. The front legs and the antennae were removed. To stop the interior of the head capsule from moving two approximately 1 mm deep incisions were made over the face of the cockroach. The vertical incision was made from top of the head between the compound eyes along the facial mid line down close to the jaws. The horizontal incision was cut below the compound eyes across the whole face. All the cut surfaces and wounds were sealed with wax. The remaining front part of the animal was fixed on the edge of a small glass plate with wax so that the upper part of the head and the compound eyes ended on top. Two small incisions were made on the dorsal corneal surface of the left compound eye so that a small sector of the cornea could be removed. To prevent dehydration the formed opening was quickly covered with an emulsion of white Vaseline and distilled water.

The cockroach preparation with the glass plate was placed on a custom made cardan arm system (courtesy of Mikko Lempeä, University of Oulu, Department of Physical Sciences; Fig. 1B) so that the left eye was placed in the centre of an imaginary sphere along whose surface the stimulation light could be moved. With the aid of a small xyz-micromanipulator (WPI, USA) sitting on top of the central pole of the cardan arm system, and below the preparation, the grease-covered opening in the eye was positioned so that the recording electrode could either be directed to the photoreceptor somata in the retina or the axons below them. The reference electrode (a chlorided silver wire) was placed through a small additional incision in the tergum (or the cornea) into the thorax (or the right eye). The recording electrode was inserted through the grease-covered opening into the

47

tissue of the left eye with a piezoelectric micromanipulator (PM 10, WPI, UK in I and II; Inchworm PZ-550, Burleigh, USA in III). The experimenter controlled the position of the head, the angle of the grease-covered opening in the eye and the direction and the location of the recording electrode by looking through a stereomicroscope (Leica, Germany in I and II; Nikon, Japan in III) equipped with cold light source, which could be freely aimed in the desired direction. A B C

Fig. 1. A. A male American cockroach. B. The custom made cardan arm system. C. The whole experimental setup.

48

Most of the recordings were done from the area of the black retina i.e. the somata of the photoreceptors. Only some of the impulse responses (those with superimposed spikes) were recorded from the photoreceptor axons (I, III). After the preparation and electrode positioning the photoreceptors were left to dark-adapt for a period of about 20-30 min. Long periods of light stimulation were always followed by dark-adaptation of at least several minutes before the next stimulation run was started. The length and level of dark adaptation was considered to be adequate when the impulse response had retained its original amplitude.

4.2 Recording and stimulation equipment and procedures

The voltage responses of the photoreceptors to light and current stimulation were recorded intracellularly with borosilicate glass capillary microelectrodes (resistances 50-150 MΩ) pulled with a Flaming/Brown puller (P-87, Sutter Instrument, USA) and usually filled with a 2 M KCl solution, whose pH was adjusted to 6.84 with a potassium-phosphate buffer. When intracellular staining and marking of the recorded cells (Stewart 1978, Stewart 1981) was desired (III), the electrodes were filled with 3% Lucifer yellow CH (Sigma Co.) in 0.1% LiCl (in this case electrode resistances were 150-250 MΩ). The electrode capacitances were compensated and the recorded signals amplified with an intracellular amplifier (SEC-1L or SEC-05L, NPI, Germany) equipped with a low voltage probe.

The light stimuli were either pulses of given intensities and lengths, or pseudorandom contrast noise sequences (French 1980b, French 1980a) with desired intensity modulations at different mean adapting background light levels. The current stimuli given through the recording electrode were either pulses of given amplitudes and lengths or pseudorandom white noise sequences with desired modulations and DC levels. In these recordings the amplifier was driven in discontinuous current clamp (switched-clamp) mode. All the stimulus voltage waveforms were originally produced, all the measurements controlled and the signals analyzed with either an Asyst (Keithley, USA) or a Matlab (Mathworks, USA) based custom made software (programming in Asyst by Eero Kouvalainen and Mikko Juusola, Biosyst in Matlab by M. Juusola) running on a computer equipped with a data acquisition card (Data Translation DT2821 or National Instruments PCI-MIO-16E-4, USA). The light stimuli were produced either with a green high intensity LED (HBG5666X, Stanley, Japan, peak wavelength 555

49

nm or B5-433-B525, Roithner Lasertechnik, Austria, 525 nm) or a stroboscope (Cathodeon Ltd, UK), whose flash was guided to the cardan arm by a quartz cable. The stroboscope was used to produce 10 μs white light flashes for eliciting impulse responses. When necessary the intensity of the stroboscope flashes could be attenuated by neutral density grey filters (Schott, Germany, transmission 49, 14, 2.5 or 0.09%). The LED was used when either longer light pulses or light intensity modulation with a background (e.g. contrast noise) was desired. The computer generated stimulus waveforms for the LED were fed to a custom made voltage-to-current driver (courtesy of Eero Kouvalainen), which controlled the current and thus the intensity modulation of the LED. The driver was equipped with a voltage output monitoring the LED current. When necessary, additional light intensity control was achieved by placing grey filters (Kodak Wratten, USA, attenuation 0.5-5.0 log units in steps of 0.5) between the stimulating LED and the eye. The monitored stimuli and the amplified response signals were low-pass filtered with a matched two-channel filter (VBF/23 or Benchmaster VBF8, Kemo, USA). A photograph of the experimental setup is shown in Fig. 1C.

When it was necessary to know the stimulus light intensity in relation to an individual photoreceptor cell and its sensitivity (I, II), i.e. while recording for example light adapted step responses, SNR, transfer and coherence functions, the light output of the LED was calibrated by counting the number of single photon responses (see e.g. Lillywhite 1977) during a prolonged dim illumination. The mean brightness of all the adapting background light levels used in the recordings was then calculated according to this calibration and the optical density of the grey filters (0 – 5.0 in steps of 0.5) used in each case. Values of absorbed photons per second (ph/s) obtained this way are so-called effective values (Juusola et al. 1994, Juusola et al. 1995) and have to be taken as measures of cell specific relative intensities only. They are not absolute values, i.e. absolute measures of absorbed photons (except at the calibration light level itself), since the quantum efficiency of the photoreceptors changes with the state of light adaptation (Smola 1976, Wu & Pak 1978, Laughlin 1981). However, the obtained stimulus intensity scaling is related both to the brightness of the stimulating light and to the sensitivity of the photoreceptor stimulated. In addition, when recording from light adapted photoreceptors, the dark adapted cells were allowed to adapt for 90-120 seconds to the adapting background i.e. mean illumination before introducing light contrast stimuli (steps or noise). This was done to ensure that the sensitivity of the photoreceptors had reached a steady-state and that most forms of adaptation were completed, including possible palisade formation and pigment migration

50

(see e.g. Butler 1973b). The light contrast (c) was defined as a change in the light intensity (ΔI) divided by the mean light background (I):

IcI

Δ= .

In the case of pseudorandom contrast modulation, the standard deviation (SD) of the stimulus was taken as ΔI (cf. Kouvalainen et al. 1994). When many different adapting background intensities were applied successively to a cell the experiments were first performed at lowest adapting backgrounds before proceeding to higher ones. In the beginning, or before frequency domain analysis, the offsets and the possible trends were always removed from the signals.

4.3 Quality of recordings

The requirements of a photoreceptor soma to be accepted for the recordings were: a sufficiently low resting potential (about -55 to -65 mV), a high input resistance (60-100 MΩ, when estimated from a response to hyperpolarizing 0.5 nA current pulse given at resting potential), and clearly visible quantum bumps at a low stimulus light level (amplitude several millivolts, cf. Fig. 1 in I). In the axon recordings only the resting potential criterion was applied. When a cell recording fulfilled these requirements, the cardan arm was moved so that the maximal amplitude for the light responses was found. All the recordings were done along this optical axis. A recording was rejected, if the sensitivity and time-courses of step responses did not return to their initial values or the above-mentioned quality criteria were not also satisfied after the recording.

4.4 Marking and anatomical identification of cells with spike responses

Some of the cells (n = 4), where impulse responses were combinations of graded responses and spike or action potential like responses, were marked and anatomically identified (III) with intracellular Lucifer yellow dye injections (Stewart 1978, Stewart 1981). In these experiments the recording electrodes were filled with 3% Lucifer yellow CH (Sigma Co.) in 0.1% LiCl and the recorded cells were marked with iontophoretic injections of the dye. Iontophoresis was made with either short 1-2 nA pulses or a constant 1 nA current and lasted all

51

together about 15 minutes. To avoid extracellular dye injections, the marked cells were tested to have similar impulse responses both before and after the dye injections. Immediately after iontophoresis the electrode was cut along the surface of the eye, the eye surgically removed from the head and fixed overnight with 10% formaldehyde in buffered insect saline. Next morning the fixed eye was embedded in Historesin (Cambridge Intruments, UK), cut into 30-80 μm slices, examined by fluorescence light microscopy (Zeiss, Germany) and photographed. When Lucifer yellow stained cell parts appeared in several slices, a camera lucida drawing combining these slices and cell parts was done by hand (Fig. 2B in III).

4.5 Localization of recorded photoreceptor responses

Electrode locations in the cockroach eyes were roughly estimated by utilizing the natural colour differences in the tissue. The part of the retina containing photoreceptor somata and pigment cells is black due to screening pigment granules and photopigments (see also Butler 1973a, Butler 1973b). In contrast, the area below the retina containing distal parts of the photoreceptor axons is brown due to pigment granules in the beginning of the axons and the rest of the axonal area up to the lamina is opalescent or very light brown, depending on the quality of incident light. These colour differences in the tissue always correlated with different impulse responses. The pure graded responses were recorded only from the photoreceptor somata on the area of black retina and the graded responses with superimposed spikes from the opalescent axon area (Fig. 2, cf. also Fig. 6 in I). In the brown area below the retina or quite proximal retina the graded responses had small spikelets on top of them.

4.6 Recordings of photoreceptor impulse and step responses in dark adaptation

Responses to very short (10 μs), bright white light flashes, i.e. impulse responses, were recorded from both photoreceptor somata and axons (I, III). All the cells were dark-adapted before the recordings. Because the inter-stimulus interval (ISI) used in these experiments was only 1 s, the more sensitive and slowly adapting photoreceptors (see II) were probably partly light-adapted during these recordings. Since there are spikes or spike-like potentials in the axons (I, III), these recordings were sampled at 2.5 kHz and not filtered.

52

Fig. 2. A schematic drawing of the cockroach eye with estimated locations of recording electrodes and different impulse responses. The figure is based on a camera lucida drawing with two intracellularly stained (here black) spiking photoreceptors (cf. Fig. 2B in III); Re = retina, Bm1 and BM2 are the two basement membranes of cockroach eye, La = lamina, Me = medulla. On the right is a micrograph of an unstained thick slice (ca. 80 μm) of fresh tissue taken with standard trans-illumination. Its rough location in the eye tissue is also shown. The colour differences in the tissue enable approximate localisation of the recording electrode and different responses (see text). Three sets of recorded light responses from different locations are shown above the drawing. Their estimated locations are marked with the microelectrode-like arrows. Each set includes impulse responses to five flashes of white light (10 μs) with different relative intensities (100; 49; 14; 2,5 and 0.09%) given at the beginning of the traces. Above the topmost set an insert shows a response with two superimposed spikes.

The stimuli used while recording the sensitivity and speed of adaptation in photoreceptors (I, II) were either 300 ms or 10 s light pulses of desired intensities with long enough ISIs (usually several seconds and with longer stimuli even minutes) for preserving a proper dark adaptation. These recordings were done only in photoreceptor somata, low-pass filtered at 700 Hz and sampled with 2 kHz. The intensity of the stimulus producing a half-maximal light response was used as a measure of sensitivity and the amplitude of depolarization at certain

53

time points after stimulus onset (300 ms and 9.9 s) in relation to the amplitude of the maximum response (A/Amax) during a bright (saturating) light pulse as a measure of adaptation speed (cf. Fig 3 in I).

4.7 Recordings of photoreceptor step responses in light adaptation

When recording step responses of light adapted photoreceptors (II), the LED output was first calibrated in each cell at a low constant intensity by counting the quantum bumps produced as a response. Together with the neutral density filters used this produced a cell-dependent intensity scale (as explained earlier in section 4.2). After this the cells were light adapted with the desired constant background intensity for 90 s and then 300 ms contrast pulses were presented on top of each background with 1 s ISIs. As before, the recording procedure advanced from the dimmest to the brightest stimuli. The contrasts used in pulses at each background were from -1 to +1 in steps of 0.2 contrast units. These recordings were filtered at 700 Hz, sampled with 2 kHz and averaged 10-20 times.

4.8 Determination of receptive field

The recordings of receptive fields were done (I) with a different setup (courtesy of E. Warrant in the University of Lund, Sweden), described in detail elsewhere (Warrant et al. 2004). In short, the setup was similar to that used in other recordings, but the light source was a Xenon arc lamp equipped with a quartz light guide and the cardan arm system was better suited for accurate receptive field measurements. The operating curves (amplitude of voltage responses vs. stimulus intensity i.e. functional sensitivity) and the horizontal receptive fields (i.e. angular sensitivity) of the dark-adapted photoreceptors were measured with impulse responses sampled at 2.5 kHz and at one degree intervals. The ISI was always long enough to maintain dark adaptation. The acceptance angles (Δρ; the width of the receptive field at 50% sensitivity level) were determined from these sensitivity functions with linear interpolations between the recorded data points.

4.9 Determination of signal-to-noise ratio

Photoreceptor SNR in the frequency domain was recorded and calculated (I) as described in detail by Kouvalainen et al. (1994, see also Juusola et al. 1994). In

54

short, when SNR was to be recorded, each cell was first individually calibrated in terms of absolute sensitivity by counting the quantum bumps produced at a very low constant illumination (as explained in section 4.2). Before starting a SNR recording session at a certain mean adapting light intensity, the cell was light-adapted for 2 min, to allow the photoreceptor to reach a steady state depolarization. During recording, a pseudorandom contrast noise sequence was repeated 20-30 times on top of each mean adapting light level. Each recorded sequence was 16 s long and the stimulus had a contrast of 0.16 or 0.32. Both the stimulus and the response were filtered at 500 Hz and sampled with 1 kHz. The time domain average of the responses gives an estimate of the signal, and any of the raw responses minus the signal is considered noise. The power spectra of the signal and the noise were calculated with an FFT algorithm from 2048 point segments (ca. 2 s) with 50% overlaps using a Blackman-Harris fourth term window (Harris 1978). The SNR in the frequency domain is the quotient of these spectra.

4.10 Determination of frequency response

Photoreceptor frequency responses i.e. transfer functions (gain and phase) were recorded and calculated (II) as described in detail by Juusola et al. (1994). In each cell, where the transfer function was recorded, a coherence function for the same responses was also calculated. Again, as in the case of the SNR, each cell was first individually calibrated in terms of absolute sensitivity by counting the quantum bumps produced at a very low constant illumination (described in detail in section 4.2). Then all other background intensities (in effective ph/s) were calculated according to this biological calibration and the optical density of the grey filters used in each case. Before starting a transfer function recording session at a certain background light intensity each cell was light-adapted for 90 s, to allow the photoreceptor to reach a steady state depolarization. Then a pseudorandom contrast noise sequence (French 1980b, French 1980a) was repeated 10-20 times on top of each adapting background light level. The recorded sequences were 16 s long and the stimulus had a contrast of 0.32. Both the stimulus and the response sequences were filtered at 500 Hz, sampled at 1 kHz and averaged in the time domain (as with SNR). The power spectra of the stimulus and the response and the cross power spectrum were calculated from this time domain averaged data with the same method as described earlier in section 4.9 (i.e. FFT, 2048 point segments, 50% overlap, a Blackman-Harris window).

55

The transfer function is the quotient of the cross power spectrum and the stimulus power spectrum. The coherence function is obtained by multiplying the cross power spectrum with its complex conjugate and dividing the result with the the multiple of the power spectra of the stimulus and the response (see e.g. Marmarelis & Marmarelis 1978, Juusola et al. 1994). From the transfer functions (gain and phase) the linear impulse responses (or 1st order Wiener kernels) could be calculated via the inverse FFT.

4.11 Evaluation of quantum bump size and shape in light adaptation

Before starting a SNR recording or pseudorandom contrast modulation (see section 4.9) a 16 second sequence of voltage response (or photon noise) to constant background light was always recorded at all the different background intensities. The power spectrum of this photon noise was calculated as described earlier (sections 4.9 and 4.10). When the phototransduction system was estimated to be a minimum phase system (cf. Wong & Knight 1980, Burton 2006) , it allowed the phase function to be calculated on the basis of the power spectrum via the Hilbert transformation. Combining the measured power spectrum and calculated phase function estimate enabled (via inverse FFT) the calculation of the average shape of the events creating the spectrum, i.e. quantum bumps (II).

4.12 Determination of impedance function

The impedance function here means a transfer function or frequency response between the injected (via the recording electrode) pseudorandomly modulated current noise stimulus and the measured membrane voltage response. This frequency response was recorded and analyzed (II) exactly as the frequency response earlier (section 4.10), only that the stimulus was current (not light) noise with a SD of 0.16 nA. A more thorough explanation of measuring the cell impedance with white noise modulated current injection is given by Weckström et al. (1992, see also Juusola & Weckström 1993).

4.13 Statistical tests

To test the normality of the distributions of the variability in light responses (I) we used both the Kolmogorov-Smirnov and the D’Agostino-Pearson test in MedCalc

56

(MedCalc Software, Belgium) and the Lilliefors test in Matlab (Mathworks, USA).

4.14 Modelling and simulations

Modelling of the cable properties of cockroach photoreceptors to test the propagation and attenuation of light responses (III) was done in the frequency domain according to the algorithm developed by van Hateren (1986), originally for fly photoreceptors and visual interneurons (van Hateren 1986, van Hateren & Laughlin 1990). The model was programmed with Mathcad (MathSoft Inc, USA) and in it the photoreceptor was modelled as a series of isopotential membrane cylinders of different sizes with connecting transfer impedances. The sizes of different cell parts were estimated on the basis of the anatomical information available on cockroach photoreceptors (Butler 1973a, Ribi 1977, Ernst & Füller 1987). In the model the current (simulating the light-induced current) was injected into the middle of the soma cylinder and the voltage transfer between the soma and the axon terminal could then be obtained.

The hypothesis concerning how a group of variable photoreceptors might work as a population was simulated with a Simulink (Mathworks, USA) program (I). A general description of this simulation is given in the summary of the results (section 5.8), and the details are given in the original paper (I) and its supplementary material. The simulation files are available from the authors via email ([email protected] or [email protected]) upon request.

57

5 Summary of the results in the original articles

In this summary the visual information processing in cockroach photoreceptors is approached from the view point of filtering the light response signals. Both in order to have a general view of the processing of light information and to dissect it into smaller pieces, this has proven to be a very fruitful and thus widely used approach also in other species, e.g. horseshoe crab Limulus (Dodge et al. 1968, Wong & Knight 1980, Wong et al. 1980, Wong et al. 1982), blowfly Calliphora (French & Järvilehto 1978, Weckström et al. 1988, Weckström et al. 1992, Juusola & Weckström 1993, Juusola et al. 1994, Juusola et al. 1995), fruitfly Drophila (Juusola & Hardie 2001a, Juusola & Hardie 2001b, Niven et al. 2003, Vähäsöyrinki et al. 2006), other Dipteran flies (French 1979, French 1980b, French 1980c, Burton et al. 2001, Burton 2006), locusts (Kuster & French 1985, Faivre & Juusola 2008), and the bee Megalopta (Frederiksen et al. 2008, Warrant 2008). Traditionally in this approach visual systems or photoreceptors are considered to be formed of a series of filters that shape the response signals and their information content, while the signals propagate along the photoreceptors or in broader sense the visual pathway. Hence, the cockroach photoreceptor together with the preceding optics can be seen as a series of filters depicted schematically in Fig. 3.

All together hundreds of adult male cockroaches were used in the experiments of this thesis. About 20 adult females were also tested and found to have similar behaviour of light responses. Intracellular recordings were made from hundreds of photoreceptors and of those around 300 cells had high enough quality, at least in some of the recorded signals, to be accepted among the results of the articles (I, II, III) and the thesis. In sections 5.1-5.6 only the properties of light responses recorded from the somata of the photoreceptors are considered, and the responses of the axons are covered in sections 5.7 and 5.8. The spectral filter of the cockroach photoreceptor (see Mote & Goldsmith 1970, Mote & Goldsmith 1971, Heimonen 1990) was not studied in the original articles of this thesis. Hence, it is not presented in this summary, but it is reviewed in the discussion part of this thesis (section 6.1). Also, the ventral, narrowing part of the eye was not studied, since all the recordings were made from either the large dorsal area or the middle of the compound eye. It is unlikely, although possible, that some properties, or at least their distributions, could be different in the ventral part of the eye.

58

Fig. 3. General scheme of filtering in cockroach photoreceptors. The incident light is filtered spatially and spectrally by the receptive field and the spectral sensitivity of the photoreceptor. The effects of these two filters on coding can be bypassed when using green light given along the optical axis. The next filter is phototranduction that produces quantum bumps, which are summed up to form receptor potentials. All the ionic currents are also filtered by the membrane of the photoreceptor. The membrane filter alone can be studied by injecting currents into the photoreceptor and recording the response voltages. The collective filtering properties of transduction and the membrane can be studied by recording the frequency response of the cell. Coherence and SNR functions between the light stimuli and voltage responses can also be determined. The additional filtering effects of the axon membrane can also be recorded. Light adaptation is known to have direct effects on at least the spatial filter and the transduction. Direct adaptational effects on the membrane filters are not known, but indirectly the changes in membrane voltage modify them.

5.1 The spatial filter

The angular sensitivity of photoreceptors forming the spatial filter was determined under dark-adapted conditions from 41 green receptors (I). The angular sensitivity showed considerable variability. The values of Δρ varied from 2.2° to 12.7°, the average being 6.4° (SD was ±2.7°) (Fig. 4; see also Fig. 4 in I). These values, like many other variable properties in the cockroach eye, were shown to be normally distributed. There was no apparent correlation of the angular sensitivity with the time-course dynamics of the light responses.

As described in the review of literature (section 2.3.3) the spatial filter narrows with the increasing light adaptation (Butler & Horridge 1973a). It is neither known nor studied here to what extent, if at all, the variability of the acceptance angle is retained in light adaptation.

59

-12 -9 -6 -3 0 3 6 9 12

0,0

0,2

0,4

0,6

0,8

1,0 S

ensi

tivity

Angle (°)

Fig. 4. The spatial filter (i.e. the size of the receptive field) showed an astonishing amount of variability between cells. Data points and fitted Gaussian curves show the widest (red, squares, Δρ = 12.7°) and the narrowest (blue, triangles, Δρ = 2.2°) horizontal dark adapted receptive fields. The average acceptance angle (Δρ) is 6.4°. The average Gaussian is shown as a dashed curve.

5.2 Properties of responses in dark and light adaptation and their variability

Before the filtering approach to cockroach photoreceptors can be continued, we need to take a more detailed look at the temporal properties of the light responses, and especially on their large variability (cf. I and II).

Like other insect photoreceptors studied, stimulation by pulses of light under otherwise dark-adapted conditions produced depolarizing voltage responses in cockroach photoreceptors. They had relatively large (up to 70 mV) peak amplitudes at the onset and a relaxation (or repolarisation) towards less depolarized potentials, when longer stimulus pulses were applied. Primarily in paper I, but also partly in II, it was shown that the variable properties of the responses included, in addition to the angular sensitivity (see the previous section), variations in absolute and relative sensitivity, adaptation speed, and signal-to-noise ratio. As with the angular sensitivity, the speed of adaptation was also found to be normally distributed.

The variability of the absolute sensitivity was not very large (about 10-fold), but in terms of the V-logI curve, it spanned over two decades, i.e. at least 100-fold differences were found (see also Heimonen 1990). In particular, the speed of

60

20 mV

2 s20 mV

200 ms

hyper-adapting

adapting

non-adapting

adaptation, that is the speed of repolarisation of constantly illuminated photoreceptors, had very large variation. For categorising purposes the cells were described with terms like “non-adapting”, “adapting” and “hyper-adapting” (Fig. 5), although the variation in relaxation speed was shown to be continuous and normally distributed both 400 ms and 9.9 s after the light onset (see Fig. 3 in I).

Fig. 5. Intracellular responses of different photoreceptors to saturating light pulses enable the cells to be categorised into three classes (see responses to 300 ms pulses on right) according to the variable time courses of their responses, although this variation actually forms a continuum. “Non-adapting” cells adapt very slowly (or in shorter timescale not at all) and “hyper-adapting” cells repolarise during illumination close to the dark resting potential. The responses to 10 s light pulses (on left) have been deliberately chosen to be similar in shape, but they repolarise to different levels of membrane voltage at constant illumination. Significantly different shapes of responses also exist (see Fig. 3A in I).

The responses to light pulses (in I and II) show prominent and variable adaptation (in terms of repolarisation during the light stimulus) as described. This behaviour was also investigated with pulses of light (contrasts between -1 and +1) under light-adapted conditions in order to quantify the contrast coding properties of the photoreceptors (II). As expected, the hyperpolarizing responses (with negative contrasts) tended to be larger than the positive ones (with positive contrasts) (Fig. 6A). To what extent this happened, and with which background light intensities, also depended on the adaptive type of the response or receptor. As in other insect species studied (e.g. Juusola 1993a, Juusola 1993b), cockroach photoreceptors show clear improvement in their contrast coding abilities at increasing background light intensities, when measured in terms of the maximum amplitude of responses to contrast pulses (Figs. 6A and 6B). However, the most noticeable finding was that this improvement saturated already with relatively low background intensities, which in our experiments were usually for effective

61

10 100 1000 10000 100000

-60

-56

-52

-48

-44

-40

mV

Intensity (ph/s)

C

B

0 mV

15 mV

3000 ph/s100 ms

10 mV

2800 ph/s

"hyper-adapting"

"non-adapting"

320000 ph/s

4 mV

5 mV

11 mV

14 mV

12 mV

100 ms10 mV

13 mV

100 ph/s

320 ph/s

1000 ph/s

3200 ph/s

32000 ph/s

"adapting"

A

intensities of 1000-10000 photons/s. The “hyper-adapting” cells, as in many other respects, formed an exception. Their responses to all contrasts between -1 and +1 were always almost non-existent (Fig. 6C). The cells categorised as “non-adapting” seemed to code the negative contrasts with a higher gain than the positive ones.

Fig. 6. Light response properties of light adapted photoreceptors show capacity for contrast coding even at low light intensities and also variation in their coding abilities. The contrasts (ΔI/I) used varied between -1 and 1, and the adjacent contrast steps were always 0.2 units apart from each other. All traces are averages of 10-20 recordings. A. Contrast response series at different backgrounds for an “adapting” cell. The background intensities are presented on the right and the corresponding “steady-state” depolarisations (in relation to the resting potential) on the left. The stimulus waveforms are presented at the bottom. B. V-logI curves for the responses in A. The maximum amplitudes at the same background light are connected with the solid lines and the “steady-state” voltages in different backgrounds with the dotted line. Notice the saturation of the “steady-state” voltage already at fairly low levels of background light intensity (around 1000 ph/s). C. Examples of a contrast response series for different cells (background intensities on the right and “steady-state” depolarisations on the left). These responses depict variation in contrast coding abilities between cells. The “non-adapting” cell only properly coded the negative contrasts and the “hyper-adapting” cell did not code any of the applied contrasts.

62

1 10 1000,0

0,1

0,2

0,3

0,4

B

Frequency (Hz)

SNR

"hyper-adapting"

1 10 1000,0

0,1

0,2

0,3

0,4

"adapting"

Frequency (Hz)

SNR

A

5.3 The SNR of contrast coding

When light-adapted, the SNR of photoreceptors of studied diurnal (e.g. Juusola et al. 1994, Juusola & Hardie 2001a, Faivre & Juusola 2008) and also nocturnal (Frederiksen et al. 2008) insect species improves considerably, although it is necessarily poor at very low light levels. Already with time-domain experiments it was clear that the cockroach SNR is fairly low at any steady-state stimulation (see previous section). When the SNR was determined with white-noise modulated light stimulation (I) by using the averaging method of Kouvalainen et al. (1994) to separate the signal and noise, it was found that in the cockroach the SNR also improves upon light adaptation (Fig. 7, see also Fig. 5 in I). However, this improvement was very limited, and the SNR always remained (at all background light intensities covered) below 1.0 in the whole frequency band, typically in the range of 0.1-0.3 (Fig. 7A). In addition, in most of the cells the improvement was even smaller and the maximum SNR attained even lower (Fig. 7B). This shows the very limited ability of cockroach photoreceptors to cope with increasing light levels, and may be related to the saturation in the improvement in contrast coding abilities that was depicted in the previous section.

Fig. 7. The SNR of contrast coding in different cockroach photoreceptors shows very limited ability to cope with increasing illumination. A. The SNR of an “adapting” cell at four different average light intensities: 100 (green), 1000 (blue), 32000 (red) and 100000 ph/s (black). “Non-adapting” (or very slowly adapting; cf. Fig. 5) cells behave similarly. B. A “hyper-adapting” cell shows very low values of SNR at all background intensities: 44 (green), 440 (blue), 4400 (red) and 14000 ph/s (black).

The poor SNR and its lack of significant increase upon light adaptation also show that the amount of information transmitted via a single photoreceptor in the

63

cockroach is very small. If the Shannon formula (e.g. Shannon & Weaver 1949) for channel capacity is applied (cf. e.g. de Ruyter van Steveninck & Laughlin 1996, Juusola & Hardie 2001a, Niven et al. 2007, Faivre & Juusola 2008, Frederiksen et al. 2008), with the very small bandwidth (ca. 15 Hz, see Fig. 5 in I) and low SNR function values (maximum about 0.3-0.4 at the highest background intensities used), we can estimate the channel capacity in the best case in bright light to be only about 70-90 bits/s. This is about one order of magnitude smaller than obtained with fast flying diurnal insects like Calliphora (Niven et al. 2007). It is also several times smaller than in other flies like Drosophila (Juusola & Hardie 2001a), and also remarkably more than twice as low as in the nocturnal bee Megalopta (Frederiksen et al. 2008). This is relevant only on the assumption that the system is linear and at steady-state, as under the experimental conditions in the present work. Generally, it can be seen from the behaviour of both the SNR functions (Fig. 7) and the coherence functions (Fig. 8) that the information capacity of cockroach photoreceptors increases with increasing ambient illumination, at least to some extent. However, since the values of coherence and SNR were generally very low, the estimates of the information capacity can only be suggestive; the exact information capacity would thus have a very wide confidence interval (Bendat & Piersol 1971). Hence, they were not calculated to any greater extent than this.

5.4 The frequency response

The dynamic linear properties of cockroach photoreceptors were estimated by determining the frequency responses, i.e. the transfer functions (gain and phase), and the coherence functions at different background intensities, using white-noise modulated light stimulation and time-domain averaging (II). The frequency responses showed a low corner frequency (around 20 Hz) and only very limited improvement of contrast gain at increasing light intensities, consistent with the experiments on contrast coding in light adaptation (cf. section 5.2). Most noticeably, there was little or no increase at all in the corner frequency with increasing light adaptation (Fig. 8A, see also Fig. 3 in II), in contrast to well studied diurnal (e.g. Juusola et al. 1994, Juusola & Hardie 2001a, Faivre & Juusola 2008) and nocturnal (Frederiksen et al. 2008) insect species. More interestingly, in the cockroach there was more variation in corner frequencies between different cells (15-25 Hz) than there ever was in a single photoreceptor between intensities, where the maximal variation encountered was 19-24 Hz.

64

Hence, it made no sense to calculate average corner frequencies or to show them as a function of intensity. Any improvements in the gain function at low frequencies were limited to light intensities between 10-1000 photons/s, above which the contrast gain was nearly constant. Consistent with the SNR determinations (Fig. 7), the values of the coherence function remained small (Figs. 8B and 8C) even after extensive averaging (10-20 times). This was especially true with the cells having a low level of “steady-state” depolarisation (< 5 mV) during prolonged light stimulation (Fig. 8C). In cells that were able to maintain a higher depolarisation level (5-15 mV), coherence values rose with increasing light intensity (Fig. 8B), and again behaved similarly to the SNR values. The higher values of coherence enabled the transfer functions to be evaluated fairly reliably in these cases (Fig. 8A). In all cells the phase part of the frequency response showed no changes at all, if we ignore the lowest intensities used (< 100 photons/s). At low intensities (< 400 ph/s) the coherence function always had values near zero and the frequency response estimates were subsequently very unreliable.

5.5 The transduction filter

To further evaluate the dynamic properties of phototransduction, the power spectra of membrane noise, induced by constant background lights of different intensity, were calculated (II). In the cells with notable levels of “steady-state” depolarisation (5-15 mV) differences in the power spectra associated with the increasing light adaptation could be found after prolonged (several minutes) light adaptation (Fig. 9A). However, these differences were also present only at low light levels, below an effective intensity of about 1000 photons/s. Above this intensity the spectra of membrane noise induced by steady light remained unchanged irrespective of light intensity. The shape and size of the events, i.e. mainly quantum bumps, producing this noise could also be evaluated, and of course these time-domain equivalents of the power spectra reproduced the results (Fig. 9B). The bump amplitude and duration seems to decrease only up to an intensity of about 1000 photons/s and then stabilises. These changes in bumps are again consistent with the findings that the contrast coding saturated at about the same intensities (1000-10000 ph/s, see Fig. 6 in section 5.2), as did the contrast gain of the photoreceptors (>1000 ph/s, see Fig. 8).

65

1 10 1000,0

0,2

0,4

0,6

0,8

1,0

Coherence

Frequency (Hz)

"adapting"

0,1

1

10

"adapting"

Phase (degrees)

Gain (mV/unit contrast)

Frequency (Hz)

-1200

-900

-600

-300

0

1 10 100

1 10 1000,0

0,2

0,4

0,6

0,8

1,0

"hyper-adapting"

Frequency (Hz)

Coherence

C

A

B

Fig. 8. Linear dynamics of photoreceptor light responses. A. Frequency response functions (gain and phase) for an “adapting” cell at six different average background intensities: 100, 320, 1000, 10000, 32000 and 100000 ph/s. The function with lowest values (green) represents the lowest background (100 ph/s) and is the only one marginally different from the others. This deviant behaviour may only exist because of the unreliability of the estimate, caused by the low values of coherence (cf. B). Although the contrast gain at low frequencies increases to some extent with increasing background intensity, the phase functions are all alike. The corner frequency of the contrast gain also shows this same temporally constant response behaviour; it is nearly invariant at about 20 Hz. B. Coherence functions for the cell in A at the same six levels of light adaptation: 100 (green), 320 (magenta), 1000 (cyan), 10000 (blue), 32000 (red) and 100000 ph/s (black). The values of coherence increase with increasing light adaptation. C. Coherence functions for a cell that always had low coherence values, i.e. for a “hyper-adapting” cell, at different levels of light adaptation: 14 (green), 140 (magenta), 440 (cyan), 1400 (blue), 4400 (red) and 14000 ph/s (black). Note that the order of traces does not systematically follow the order of increasing intensities. This even further implies that the cells with low “steady-state” depolarisation levels fail to light adapt. The frequency response estimates of these “hyper-adapting” cells would have been unreliable because of the always low coherence values, and thus they were not calculated.

66

0 100 200

0,0

0,1

0,2

Relative voltage

Time (ms)C D

B 0 100 200

0,0

0,1

0,2

Time (ms)

Relative voltage

1 10 100

10-3

10-2

10-1

100

101

"adapting"

Frequency (Hz)

Power density (mV2/Hz)

1 10 100

10-3

10-2

10-1

100

101

"hyper-adapting"

Power density (mV2/Hz)

Frequency (Hz)

A

.

Fig. 9. Transduction filter, i.e. the relationship between the background or adapting light intensity and the properties of quantum bumps or photon noise induced by constant illumination. In different photoreceptors, the power spectra of photon noise induced by constant background light demonstrated different quantum bump behaviour as a function of light intensity. A. In the “adapting” cells there were some changes in the power spectra at the lowest background intensities: 10 (blue), 100 (blue) and 1000 ph/s (red). The brighter backgrounds (10000 and 100000 ph/s; black) led to similar photon noise and spectra as 1000 ph/s without any noticeable changes with increasing intensity of illumination. B. Estimates of average quantum bumps for the “adapting” cell in A at four different light intensities: 10 (green), 100 (blue), 1000 (red) and 100000 ph/s (black). The estimates are calculated on the basis of the photon noise spectra in A and drawn with the corresponding colours. Notice that between the intensities of 1000 and 100000 ph/s, the bump estimates are almost identical. C. In “hyper-adapting” cells the power spectra were all alike in spite of the background intensity: here 44 (green), 440, 1400, 4400 and 14000 ph/s. The “steady-state” depolarisation of these cells was at or slightly above (here 1 mV) the resting potential, irrespective of the light intensity. D. Estimates of average quantum bumps for the “hyper-adapting” cell in C at two different light intensities: 44 (green) and 14000 ph/s. The estimates are calculated on the basis of the photon noise spectra in C, and just as the spectra, they are very similar.

67

The “hyper-adapting” cells, i.e. cells that had repolarised to the level of resting potential during the light adaptation period before the recordings, didn’t show any changes in the power spectra of constant-light-induced noise as a function of increasing light intensity (Figs. 9C and 9D). This is probably simply due to the fact that there seems to be almost no light response remaining in these cells after the preceding light adaptation period, irrespective of the intensity; only some sporadic discrete events resembling single quantum bumps (see Fig. 3A in I or Fig. 1B in II).

5.6 The membrane filter

The properties of the non-transductive part of the cell membrane were investigated with injections of current (II). Both current pulses and white noise modulated current were used. Hyperpolarising current pulse injections yielded estimates for the input resistances of the photoreceptors in the range of 60-150 MΩ, with time constants in the range of 20-50 ms (Fig. 10A). Whereas hyperpolarizing current pulses induced more or less exponential charging of the cell membrane in the dark, the depolarizing pulses showed clear rectifying behaviour of the membrane voltage, indicating the presence of a strong voltage-dependent K+ conductance, as in all other studied insect species (e.g. Weckström et al. 1991, Laughlin & Weckström 1993, Weckström & Laughlin 1995).

The determination of membrane impedance mirrors the results obtained with current pulses. In the dark and below the resting potential the impedance was large (100±20 MΩ) for frequencies below about 10 Hz, and the high-frequency roll-off showed the conspicuous behaviour of a single-stage RC-filter (Fig. 10B). The corner frequency was 8-10 Hz depending on the cell. Already at resting potential, and even more so above that, the membrane impedance had more complex behaviour, because of the voltage-dependence. The corner frequency of the resting membrane was about 20 Hz and that of the depolarized (ca. 10 mV) membrane around 50 Hz. No correlation was found with the impedance functions and the variable functional properties of the photoreceptors. The shape and general behaviour of impedance functions were very similar in all the cells studied.

68

0 100 200 300 400 500 600

-60

-50

-40

-30

-20

-10

0

10

-0,5

0,0

0,5

Time (ms)

Voltage (mV)

Current (nA)

1 10 100

10

100

Frequency (Hz)

Impedance (MΩ)

B

A

Fig. 10. Membrane properties of photoreceptors, i.e. the membrane filter. A. Typical current clamp recording showing the responses of the photoreceptor membrane (upper panel) to current pulses (lower panel). The resting potential of the cell (here 0 mV) was -62 mV. B. Typical impedance functions for the photoreceptor cell membrane at different voltages. The average recording voltages in relation to the resting potential (-61 mV) were -25 mV (green), 0 mV (blue), +8 mV (red) and +10 mV (black). The depolarisations were induced either by constant illumination (+8 mV, red) or by current injection (+10 mV, black).

5.7 The “axon filter” and spike coding

Although the voltage responses of cockroach photoreceptors to light, especially impulse responses, are superficially very similar to those found in other insect species studied, the photoreceptors were found to be able to generate action potentials (or spikes) in their axons (I, III). The spikes are occasionally seen in recordings when the electrode tip is located in the retina, but then only as small peaks of a few millivolts on top of the rising phase of the impulse response (Fig. 2 on page 50; see also Fig. 1 in III). However, when recordings from photoreceptor axons were made, it was clear that the signalling mode changes in the axons. Both graded impulse responses and superimposed action potentials are transmitted. Occasionally several repetitive spikes were observed on top of one impulse response (Fig. 1C in III and Fig. 6 in I). The spikes could also be triggered by depolarizing current injections (Fig. 3 in III). Via intracellular staining these cells

69

with graded responses and superimposed spikes were shown to be photoreceptors (Fig. 2; see also Fig. 2B in III). The spike responses are fairly small, maximally about 10-20 mV (measured from the top of graded response). The function of the spiking signals was explained with simplified cable modelling (van Hateren 1986, van Hateren & Laughlin 1990), with which it could be demonstrated that after arriving at the distant (up to 1-1,5 mm) presynaptic terminals of photoreceptor axons in the lamina or medulla, the graded potential is for the most part attenuated.

5.8 A hypothesis concerning the function of the cockroach visual system

Based on all the functional variability found in the photoreceptors, and their low SNR and small channel capacity even when light adapted (I), and the proposed spike coding in the axons (I, III), a hypothetical model was formulated (I) to simulate the possible function of the cockroach visual system. Together with the functional variability, the other foundation of the hypothesis was provided by previous anatomical findings (Ribi 1977, Ernst & Füller 1987), which show a significant, but variable amount of pooling of photoreceptor axons (6-20) before synapse formation in lamina. In addition there are large presynaptic arborisations in the 2nd order neurons, whose number, on the other hand, is small in comparison to the large number of photoreceptors.

According to our hypothesis, a group of photoreceptors (with their functional variability and action potential coding in the axons), when pooled onto the same 2nd order cell, could perform better than a system of photoreceptors with identical properties. The hypothesis was tested with a simplified computer program, where the simulations of 12 pooled photoreceptors seemed to support the hypothesis. In the comparison, even if the identical cells were all of the “best” type (with highest SNR and coherence) recorded in the experiments, the group of variable photoreceptors performed better, i.e. produced a more reliable input to the 2nd order cell.

70

6 Discussion

In this discussion only the main findings of the original articles (I, II, III) are discussed with respect to vision in dim light. A broader (and in several respects more detailed) discussion can be found in the original articles themselves.

6.1 Functional variability of the photoreceptors

The large functional variability found in this study (I, II) in many respects corresponds to structural variability found earlier (Butler 1971, Butler 1973a, Butler 1973b, Ribi 1977, Ernst & Füller 1987, Füller et al. 1989, Mote 1990, Ferrell & Reitcheck 1993). Variability is in general highly unusual in visual systems, where a high level of regularity is thought to be of great importance in order to form a reliable picture of the surroundings (e.g. French et al. 1977, Stavenga 1979, Laughlin 1981, Land 1985). According to our results (I, II), the variability in the function of cockroach photoreceptors seems to be achieved via structural variations, variations in phototransduction, and variations in light adaptation. In contrast, as discussed in paper I, and reported in the findings of paper II, the membrane filtering properties of photoreceptors are always nearly identical. Hence, variability can’t be explained through differences in the non-tranductive membrane.

The structural variation in lens and rhabdom sizes etc. (Butler 1971, Butler 1973a, Butler 1973b, Ferrell & Reitcheck 1993) probably explains most of the variability in the angular sensitivity (cf. e.g. Snyder & Horridge 1972, Snyder et al. 1973) and maybe also, at least partly, the variations in the overall sensitivity (cf. e.g. Snyder et al. 1973, Howard et al. 1987, Roebroek & Stavenga 1990a) of the photoreceptors. The variations reported between ommatidia in the formation of the palisade and simultaneous migration of pigment granules (Ferrell & Reitcheck 1993) can (and probably will) cause variations in photomechanical adaptation. These can partly explain the variations in the speed of long term adaptation.

On the other hand, the variations in phototransduction and its adaptational properties are difficult to explain and even separate from each other. Part of the variation is probably embedded directly in the processes of photochemical adaptation, which takes place either inside, and/or in interaction with, the processes of phototransduction. One possible explanation for the existence of hyper-adaptation can be given. It has been reported (Krause et al. 2008a) that the

71

light-gated channels in the cockroach photoreceptors are most probably only of the TRPL-type (transient receptor potential like). In contrast, in Drosophila they are mostly of the TRP-type (transient receptor potential) with a smaller portion of the TRPL-type (reviewed by Hardie & Raghu 2001, Hardie 2006, Minke & Parnas 2006). It is known that in light adaptation TRPL-channels are translocated out of the rhabdomeric microvilli (Bähner et al. 2002, Meyer et al. 2006). If this also happens in the cockroach, it would of course lead to a hyper-adaptation, where light responses disappear. TRP-mutants (lacking TRP-channels) of Drosophila were originally named because they behaved in light like the “hyper-adapting” cells of cockroaches, i.e. they had only transient receptor potentials (reviewed by Hardie 2006). Their receptor potentials are naturally caused only by currents through TRPL-channels.

The possible translocation of TRPL-channels from cockroach photoreceptors might largely explain all the variations in the speed of adaptation. In article II it was reported that all the cockroach photoreceptors adapt or repolarise close to the resting potential, when exposed to prolonged (several minutes) illumination. Even the cells that have the highest values of SNR (about 30% of all photoreceptors) have a “steady-state” depolarisation of only 10-15 mV after a few minutes of constant bright light (equivalent to hundreds of thousands of effective photons/s). The rest of the cells repolarise in the same situation even to a lower value of depolarisation, most of them within a few millivolts of the resting potential. Even the former case is not really steady-state, because after tens of minutes of constant illumination all cells have repolarised close to rest. Time to resting potential in bright constant light varies from about 400 ms to about 40 min between different cells. So, in this respect all photoreceptors of the cockroach are “hyper-adapting”, only with very variable time scales. If the translocation times of TRPL-channels can also vary this much, this might explain the variations in speed of adaptation. This might also explain why cockroaches prefer dim environments and avoid long lasting bright illumination (cf. section 2.3.1) – they become blinded. There is one finding that speaks against this above depicted TRPL-hypothesis, and that is the presence of quantum bump-like phenomena during the repolarization phase of the rapidly adapting photoreceptor responses (see Fig 3A in I and Fig. 1B in II). If the translocation were responsible for the “hyper-adaptation”, one would expect the phototransduction gain to decrease uniformly. The presence of quantum bumps suggests that most of the microvilli are shut off from the transduction process, and only those remaining are still generating bumps. But surely, this shutting-off process could be based on the translocation.

72

Variation in spectral sensitivity

One way to further increase the variability of the photoreceptors - and thus according to our hypothesis on population coding (I), the reliability of the second order neurons - would be to have variation in the spectral characteristics of the cells. Mote and Goldsmith (1970, 1971) made their spectral sensitivity recordings with white-eyed mutant cockroaches and did not report any variability in the sensitivity curves. Instead, Heimonen (1990) recorded spectral sensitivities of wild type cockroaches and found considerable variability between photoreceptors in their spectral properties (Fig. 11).

Fig. 11. Variability in the spectral sensitivity curves of wild type Periplaneta. A. Spectral sensitivity curves of eight green receptors show variability in the wavelength of the peak sensitivity and also some variation in their general shape. B. Spectral sensitivity of UV receptors. C. Photoreceptors having somewhat peculiar spectral sensitivity curves with three peaks and maximum sensitivities in the UV range were also encountered more than once. D. The average spectral sensitivity curve based on all the curves presented in A, B and C shows flatness of sensitivity over the whole visible spectrum. (Modified from Heimonen 1990.)

The peaks of spectral sensitivity varied both in green (Fig. 11A) and UV receptors (Fig. 11B). The overall shape of the spectral sensitivity curves also varied and a couple of times a cell with a three-peaked sensitivity curve was recorded (Fig.

73

11C). It is not being suggested here that these spectrally different photoreceptors in the cockroach have different photo- or screening pigments, although this is not completely unknown. For instance this is reported in Mysis shrimps adapted to different visual environments (Jokela-Määttä et al. 2005, Pahlberg et al. 2005). However, this would require the necessary genetic machinery to produce these pigment variations and this is unlikely to exist in the cockroach. Instead, it is suggested that the variable structure of individual photoreceptors inside cockroach ommatidia (Butler 1971) is the primary source of this spectral variation. Photoreceptors with different sizes, shapes and locations of light sensitive rhabdomeres unavoidably act as spectral screening filters to each other (Snyder et al. 1973). Even if the rhabdomeres in a fused rhabdom are similar in shape, they act as lateral screening filters to each other. In cockroaches this filtering has to be enhanced, since some of the cells having distal rhabdomeres (Butler 1971) modify the spectral properties of the incident light before more proximal rhabdomeres have a chance to absorb it. This leads to different spectral sensitivity curves in different photoreceptors, even though they have exactly the same photopigments. Of course UV receptors and green receptors (Mote & Goldsmith 1970, Mote & Goldsmith 1971) have different rhodopsins, but they also unavoidably modify each other’s spectral sensitivities due to this lateral spectral filtering (Snyder et al. 1973).

What would be the possible benefits of this spectral variation for cockroach vision? To start with there is considerable confusion in the literature. Butler (1971) found three UV cells and five green receptors in each ommatidium. Mote (1990) claimed that all the photoreceptors synapsing directly into the medulla are UV cells. However, the anatomical studies (Ribi 1977, Ernst & Füller 1987) show that there is only one photoreceptor per ommatidium terminating in the medulla. This would leave two UV cells per ommatidium to terminate in the lamina. The structure of the lamina is so irregular and the bundling of photoreceptor axons so random that it is difficult to imagine UV and green receptors synapsing separately and orderly onto their own 2nd order neurons. This leads one to think that they are pooled onto the same monopolar cells in the lamina. The gain for vision would be increased variability and thus improved reliability at the level of the 2nd order cells, and also an enhanced photon catch over a wider spectral range, since the overall spectral sensitivity of randomly pooled photoreceptors would be nearly flat (Fig. 11D). This arrangement would still leave room for the reported (Mote & Rubin 1981, Kelly & Mote 1990b, Mote 1990) UV sensitive cells in the optic ganglia: some of the UV receptors could still terminate directly in the medulla.

74

6.2 Is functional variability of photoreceptors optimization to low light levels?

Both previous and present findings show that there are large and uncommon structural irregularities in the compound eyes of Periplaneta (see section 2.3.2) and similarly large variability in the functional properties of its photoreceptors (I, II). Following discussions with colleagues at meetings a possibility has arisen that these variations could have evolved as degenerative mutations during a long life in captivity and laboratory breeding (cockroaches are very popular experimental animals: Huber et al. 1990a, Huber et al. 1990b). This is probably not so, since during of our experiments we have used hundreds of cockroaches, if not thousands, from several different colonies from commercial breeders and science laboratories in different places and the results always include this high variability. In addition, it has lately been reported that another large dark active cockroach species (Gromphadorhina portentosa) has similar irregularities in the structure of its compound eyes (Mishra & Meyer-Rochow 2008), but this species has been living in laboratory conditions only for a short while. As a conclusion, these variations are most probably produced by natural evolution and they are likely to be an adaptation to something.

The behaviour of Periplaneta (see section 2.3.1) strongly supports the possibility that cockroaches have evolved a dark-active and nocturnal lifestyle during their long evolution (Kambhampati 1995), and that it has proven to be a very succesfull strategy (e.g. Bell & Adiyodi 1982, Bell 1990, Bell et al. 2007). The cockroach visual system may be a secondary sensory information source, when compared to mechano- and chemosensory systems (e.g. Seelinger 1990, Zill 1990, Mizunami et al. 1998, Ye et al. 2003, Willis & Avondet 2005, Okada & Toh 2006), but nevertheless it is still present in the form of large compound eyes and ocelli (e.g. Butler 1973a, Mote 1990, Mizunami 1994).

Since sensory systems, especially visual systems, are known to have evolved to be comparatively well optimized to the ecological niche (e.g. Laughlin 1989, Laughlin 1990, Weckström & Laughlin 1995, Laughlin 1996, Warrant 1999, Warrant 2004), it can be concluded that the vision of the cockroach is probably as equally well adapted. The irregularity and variability are so dominant features of cockroach vision that they too must be ecologically adapted. Our simulations (I), which basically combine all the known variability and functional features of the cockroach visual system, support the idea of optimization to dim light conditions. They do this by showing that spatial pooling of several of these randomly variable

75

photoreceptors into one 2nd order cell actually enables the visual system to see more reliably than it could see by pooling an equal number of identical cells. This is true for all types of photoreceptors encountered in our experiments. In fact our simulations also showed (unpublished) that the addition of further variability still further increases the reliability. This supports the possibility that variations in the spectral sensitivity (see previous section) could also be utilised for this purpose.

However, simulations and modelling, even though useful, can only prove what is possible. It is surely self-evident that before it can be said that the cockroach visual system really works as suggested by the simulations, the actual intracellular recordings from presynaptic photoreceptor axons and 2nd order neurons are necessary. With the current methods this has up to now been unsuccessful and an improved preparation technique to enable insertion of a recording electrode directly into the lamina is under development.

6.3 Is filtering of visual information in cockroach photoreceptors matched to low levels of luminance?

Only about 30% of the photoreceptors studied here were suitable for reliable frequency response determinations, since all the others had coherence values below 0.5 at all the background intensities used (even with time-domain averaging). These low values of coherence correlated with the low values of “steady-state” depolarisations. In these cells the frequency response estimates were always so unreliable, that they are not shown in the figures and were omitted from all the descriptions of frequency response features.

Nevertheless, all the features of cockroach photoreceptors determined showed similar temporal features, and almost independent of the light intensity used (II). The corner frequencies in the transfer functions (above 300 ph/s between 15-25 Hz), constant-light-induced noise power spectra (above 100 ph/s ca. 20 Hz) and impedance functions (ca. 20 Hz in resting membrane) were all almost exactly matched to each other. The SNR and coherence functions also had their highest values only below about these same frequencies. This means the speed of phototransduction and the properties of the cell membrane seem to be matched.

However, the corner frequency of the membrane impedance rises (up to 50-60 Hz) when the membrane is depolarised (Fig. 10B), probably due to increased activation of voltage-dependent K+ channels. This means that in those conditions the membrane would allow faster voltage changes than what was recorded in our experiments. It doesn’t make any sense to enable this without a good reason,

76

since opening more channels to lower the impedance and to raise the corner frequency costs energy in the form of increased ionic currents (Laughlin et al. 1998, Laughlin 2001, Niven et al. 2007, Niven & Laughlin 2008). These ions then have to pumped back to maintain the homeostatic concentrations, and this is an energy consuming process. This allows us to speculate that in transient responses, during the first second, or even firsts tens of seconds, after a change in light intensity, the dynamics of the light responses might be better than determined here. This would occur when the level of depolarisation in the light responses, as it was in many cells (see Fig. 5), is markedly higher (> 15 mV) than it was during all our linear dynamics experiments. This would enable higher information capacities during these transient phases of the responses. The dynamics of the transient responses can’t be captured with white noise analysis and new kinds of methods need to be developed to study this.

To summarize, all the filters present in cockroach photoreceptors are more or less matched to each other’s speeds and they all have fairly low corner frequencies, even in bright light, especially when compared to diurnal species (Juusola et al. 1994, Juusola & Hardie 2001a). However, the photoreceptors of locusts (Faivre & Juusola 2008) and the nocturnal bee Megalopta (Frederiksen et al. 2008) are about as slow as in the cockroach. Nevertheless, it can be said that temporally, cockroach vision is matched to low light conditions. In addition, in most respects (quantum bump size and shape, contrast gain, corner frequencies etc.) cockroach photoreceptors already reach their peak performance at very low light intensities (hundreds or thousands of photons/s). When these findings are combined, it can be said, that the cockroach visual system appears to have been optimized to cope with low light conditions. Moreover, the inability to improve performance in any significant manner in bright light can be seen as a permanent optimization of vision in dim light.

6.4 Is the visual information spike coded in the cockroach photoreceptor axons?

Usually photoreceptor axons, also in compound eyes, are relatively short (e.g. Järvilehto & Zettler 1973, Ribi 1979, Ribi 1981, Greiner et al. 2004b) and light information is generally thought to propagate along them in the form of graded responses (see e.g. Zettler & Järvilehto 1973, Järvilehto 1978, Weckström 1987). The axons of cockroach photoreceptors are known to be several times longer than what is normal in compound eyes (see section 2.3.2). The lamina is usually

77

situated almost immediately below the retina and the basement membrane, but in the cockroach it is far away from them (see Fig. 2).

In this thesis it was found that the axons of cockroach photoreceptors seem to utilise spike coding to propagate the signals to the distant lamina (I, III). Action potentials have also been recorded from other photoreceptors of insect compound eye, namely in the eyes of the honeybee drone (Baumann 1968). However, in drones it has been shown that fully fledged action potentials are generated only as responses to sub-saturating or saturating light stimuli. In physiological conditions the voltage-gated sodium channels responsible for the spikes in drone photoreceptors are used to amplify the small membrane voltage modulations that are generated by the queen bee flying against a bright sky background (Coles & Schneider-Picard 1989, Vallet & Coles 1991, Vallet et al. 1992, Vallet & Coles 1993). The spikes in the axons of cockroach photoreceptors are different. They are generated as a response to graded receptor potentials as low in amplitude as 5-10 mV (I), which means that when dark adapted they might be generated even by individual quantum bumps. They are small in size (the largest recorded being only 10-20 mV) when measured from the top of the graded response they were superimposed upon. This can be interpreted in two ways: either the action potential generation site is (electrically) far away from the site of the recording, or the intracellular recording electrode causes significant damage to the membrane and induces a large leak. And of course there is also the third option that the spikes really are that small. The question as to which of these possibilities is the right one, can only be solved beyond dispute by further recordings from the axons.

However, with the aid of compartmental Hodgkin-Huxley type model of cockroach photoreceptors (Salmela et al. 2004) one can show that the generation site of fully fledged action potentials could be proximal to the second basement membrane. This would, according to this modelling, be (electrically) far enough to attenuate the spikes roughly to the same extent as in our experimental data in papers I and III. In addition, this same study, and a previous study (Heimonen et al. 2001) with a further improved version of the model in paper III, re-estimated the attenuation of graded responses while they propagate from the soma to the distant lamina. It was evaluated that the attenuation is even greater than what was estimated in article III. Only 10-20% of the amplitude of graded responses was present, and the higher response frequencies even further attenuated, when the graded responses reached the synaptic terminal. This further emphasizes the need for spike coding in the axons.

78

The temporal properties of spike coding were not systematically studied in any of the original articles in this thesis, since the axon recordings were generally not stable and sufficiently long lasting for this purpose. It was earlier reported by Heimonen (1999) that from 15 successful recording sessions from photoreceptor axons it was possible only twice to quickly record responses to repeated light flashes with different frequencies (Fig. 12).

Fig. 12. Impulse responses from a cockroach photoreceptor axon to flashes of light repeated with different frequencies. The frequencies used in each trace were from top to bottom 5, 8 and 10 Hz. (Heimonen 1999).

In both cells the action potentials were not generated at all when the frequency of the flashes was raised above 10 Hz, and they were found in less than a 1:10 relation when stimulating at 10 Hz. Instead, an action potential was generated by almost every flash at a frequency of 8 Hz, and always below that frequency. This gives only a rough estimate of the temporal properties of spike coding. On the other hand, in those recordings where more than one spike per impulse response were found (Fig. 2, see also Fig. 6 in I), the maximal frequency of spikes on top of one impulse response was estimated to be about 40-50 Hz. Noting more is known concerning the temporal properties of spike coding in the photoreceptor axons of the cockroach.

79

6.5 Are cockroach photoreceptors optimized for vision in dim light?

Even though cockroach compound eyes have large numbers of photoreceptors (Walther 1958b, Wolken 1971, Nowel 1981, Stark & Mote 1981, Füller et al. 1989), it is argued here that even in bright light the spatial resolution of cockroach vision is always fairly poor (I). When light-adapted, the spatial resolution of individual photoreceptors has been shown to sharpen (Butler & Horridge 1973a) due to palisade formation and intracellular pigment migration, i.e. pupil dilation (Snyder & Horridge 1972). But this is probably only a side product of overall sensitivity regulation (e.g. Snyder et al. 1973, Howard et al. 1987, Roebroek & Stavenga 1990a), necessary to maintain the functionality of photoreceptors even in bright environments via decreasing the photon catch. However, primarily because of the likelihood of fairly extensive spatial summation at the first visual synapse of cockroach compound eyes (Ribi 1977, Ernst & Füller 1987), the spatial resolution is inevitably low in all illumination conditions. This kind of strategy of vision can be seen as a permanent optimization favouring dark or dim-light conditions (I).

It is also argued here that the temporal resolution of cockroach vision is always, even in bright light, fairly poor (II). The temporal resolution of all compound eyes is fairly low in the dark-adapted state (e.g. Howard et al. 1984, Laughlin & Weckström 1993, Juusola et al. 1994, Juusola & Hardie 2001a, Frederiksen et al. 2008), because of the natural need for temporal summation to make sense of inevitably noisy signals in dim light (e.g. van Hateren 1992, Warrant 1999, Warrant 2006). It is shown here that cockroach photoreceptors, unlike other insect visual systems that have been studied (e.g. Howard et al. 1984, Laughlin & Weckström 1993, Juusola et al. 1994, Juusola & Hardie 2001a, Frederiksen et al. 2008), fail to speed up with increasing light adaptation (II). Even in bright light, both temporal resolution and contrast coding abilities stay at about the same level as they are at intensities of about 1000 photons/s. On the other hand, if there is no need for faster visual performance in any conditions, this can be seen as a way to save energy (Laughlin et al. 1998, Laughlin 2001, Niven et al. 2007, Niven & Laughlin 2008). Faster vision would need faster phototransduction, i.e. more and smaller quantum bumps, and thus larger amounts of activated microvilli in the rhabdomeres. It would also mean larger numbers of opened voltage-gated channels to make the cell membrane faster. Both of these would lead to an increase in gross ion flux through the membrane, and pumping ions back to maintain homeostasis would cost energy. Hence, the permanently

80

slow vision of the cockroach (II) can be seen as a permanent optimization favouring dark or dim-light conditions.

81

7 Conclusions

The processing of visual information in the cockroach Periplaneta americana is shown to be deviant in many respects from known strategies of neural information processing in other superficially similar visual systems. All the uncommon functional features of cockroach photoreceptors found in this thesis can be seen as adaptations or optimisations to vision in dim light.

1. The large variability in the functional properties of cockroach photoreceptors, when combined with a fair amount of neural pooling in the first visual synapse, enables the cockroach visual system to function more reliably than with functionally identical cells.

2. The temporal dynamics and light information filtering properties of cockroach photoreceptors seem to be permanently matched or optimised to low levels of illumination. The information processing capabilities of all types of cockroach photoreceptors were, at all levels of light adaptation (even in bright light), very similar to processing capabilities in dim light, i.e. equipped with slow dynamics and low SNR.

3. Spike coding is introduced in cockroach visual processing in an astonishingly early phase, i.e. already in the photoreceptor axons. When combined with functional variability and neural pooling in the first visual synapse this feature on its own further increases the reliability of the cockroach visual system in dim environments.

The cockroach visual system, at least when judged on the basis of the function of the photoreceptors, seems to be permanently optimised to operate in dim light: the functional capacity is no better in bright light than at very low levels of ambient illumination.

82

References Autrum H (1958) Electrophysiological analysis of the visual systems in insects.

Exp Cell Res 5: 426-439. Autrum H (1981) Light and dark adaptation in invertebrates. In: H. Autrum (ed)

Handbook of sensory physiology. Berlin, Heidelberg, New York, Springer-Verlag Vol. VII/6C, p 1-91.

Autrum H & Stöcker M (1952) Über optische Verschmelzungsfrequenzen und stroboskopisches Sehen bei Insekten. Biol Zentralbl 71: 129-152.

Ball JB (1971) The receptor site for photic entrainment of circadian activity rhythms in the cockroach Periplaneta americana. Ann Ent Soc Am 64: 1010-1015.

Baumann F (1968) Slow and spike potentials recorded from retinula cells of the honeybee drone in response to light. J Gen Physiol 52: 855-875.

Bell WJ (1990) Biology of the cockroach. In: I. Huber, E. P. Masler and B. R. Rao (eds) Cockroaches as models for neurobiology: applications in biomedical research. Boca Raton, Florida, CRC Press, Inc. Vol. I, p 7-12.

Bell WJ & Adiyodi KG (eds) (1982). The American cockroach. London, Chapman and Hall.

Bell WJ, Roth LM, Nalepa CA & Wilson EO (2007) Cockroaches: ecology, behavior, and natural history. The Johns Hopkins University Press, Baltimore.

Bendat JS & Piersol AG (1971) Random data: analysis and measurement procedures. John Wiley & Sons, Inc., New York.

Bhandawat V, Reisert J & Yau KW (2005) Elementary response of olfactory receptor neurons to odorants. Science 308: 1931-1934.

Briscoe AD & Chittka L (2001) The evolution of color vision in insects. Annu Rev Entomol 46: 471-510.

Burdohan JA & Comer CM (1996) Cellular organization of an antennal mechanosensory pathway in the cockroach, Periplaneta americana. J Neurosci 16: 5830-5843.

Burkhardt D, de la Motte I & Seitz G (1966) Physiological optics of the compound eye of the blowfly. In: C. G. Bernhard (ed) The functional organization of the compound eye. Oxford, Pergamon Press, p 51-62.

Burns ME & Baylor DA (2001) Activation, deactivation, and adaptation in vertebrate photoreceptor cells. Annu Rev Neurosci 24: 779-805.

Burton BG (2006) Adaptation of single photon responses in photoreceptors of the housefly, Musca domestica: a novel spectral analysis. Vision Res 46: 622-635.

Burton BG, Tatler BW & Laughlin SB (2001) Variations in photoreceptor response dynamics across the fly retina. J Neurophysiol 86: 950-960.

Butler R (1971) The identification and mapping of spectral cell types in the retina of Periplaneta americana. Z vergl Physiol 72: 67-80.

83

Butler R (1973a) The anatomy of the compound eye of Periplaneta americana L. 1. General features. J Comp Physiol 83: 223-238.

Butler R (1973b) The anatomy of the compound eye of Periplaneta americana L. 2. Fine structure. J Comp Physiol 83: 239-262.

Butler R & Horridge GA (1973a) The electrophysiology of the retina of Periplaneta americana L. 1. Changes in receptor acuity upon light/dark adaptation. J Comp Physiol 83: 263–278.

Butler R & Horridge GA (1973b) The electrophysiology of the retina of Periplaneta americana L. 2. Receptor sensitivity and polarized light sensitivity. J Comp Physiol 83: 279–288.

Bähner M, Frechter S, Da Silva N, Minke B, Paulsen R & Huber A (2002) Light-regulated subcellular translocation of Drosophila TRPL channels induces long-term adaptation and modifies the light-induced current. Neuron 34: 83-93.

Camhi JM (1980) The escape system of the cockroach. Sci Am 243: 144-156. Camhi JM & Johnson EN (1999) High-frequency steering maneuvers mediated

by tactile cues: antennal wall-following in the cockroach. J Exp Biol 202: 631-643.

Camhi JM & Levy A (1989) The code for stimulus direction in a cell assembly in the cockroach. J Comp Physiol A 165: 83-97.

Coles JA & Schneider-Picard G (1989) Amplification of small signals by voltage-gated sodium channels in drone photoreceptors. J Comp Physiol A 165: 109-118.

Comer CM, Parks L, Halvorsen MB & Breese-Terteling A (2003) The antennal system and cockroach evasive behavior. II. Stimulus identification and localization are separable antennal functions. J Comp Physiol A 189: 97-103.

Cooter RJ (1973) Visual and multimodal interneurones in the ventral nerve cord of the cockroach, Periplaneta americana. J Exp Biol 59: 675-696.

Cowan NJ, Lee J & Full RJ (2006) Task-level control of rapid wall following in the American cockroach. J Exp Biol 209: 1617-1629.

Dacke M, Byrne MJ, Scholtz CH & Warrant EJ (2004) Lunar orientation in a beetle. Proc R Soc Lond B Biol Sci 271: 361-365.

Dacke M, Nilsson DE, Scholtz CH, Byrne M & Warrant EJ (2003) Animal behaviour: insect orientation to polarized moonlight. Nature 424: 33.

Dayan P & Abbott LF (2005) Theoretical Neuroscience: Computational and Mathematical Modeling of Neural Systems. MIT Press, Cambridge, Massachusetts.

de Ruyter van Steveninck RR & Laughlin SB (1996) The rate of information transfer at graded-potential synapses. Nature 379: 642-645.

Dodge FA, Jr., Knight BW & Toyoda J (1968) Voltage noise in Limulus visual cells. Science 160: 88-90.

84

Edwards DH (1982a) The Cockroach DCMD Neurone I. Lateral Inhibition and the Effects of Light-and Dark-adaptation. J Exp Biol 99: 61-90.

Edwards DH (1982b) The Cockroach DCMD Neurone II. Dynamics of Response Habituation and Convergence of Spectral Inputs. J Exp Biol 99: 91-107.

Egelhaaf M (2006) The neural computation of visual motion information. In: E. J. Warrant and D.-E. Nilsson (eds) Invertebrate vision. Cambridge, Gambridge University Press, p 399–461.

Egelhaaf M & Kern R (2002) Vision in flying insects. Curr Opin Neurobiol 12(6): 699-706.

Egelhaaf M, Kern R, Krapp HG, Kretzberg J, Kurtz R & Warzecha AK (2002) Neural encoding of behaviourally relevant visual-motion information in the fly. Trends Neurosci 25: 96-102.

Egelhaaf M & Warzecha AK (1999) Encoding of motion in real time by the fly visual system. Curr Opin Neurobiol 9(4): 454-60.

Ernst A & Füller H (1987) Zur Feinstruktur der Lamina Ganglionaris von Periplaneta americana (L.). Zool Jahrb Anat 115: 393-416.

Faivre O & Juusola M (2008) Visual coding in locust photoreceptors. PLoS ONE 3: e2173.

Ferrell BR & Reitcheck BG (1993) Circadian changes in cockroach ommatidial structure. J Comp Physiol A 173: 549-555.

Frederiksen R (2008) Ommatidial adaptations for vision in nocturnal insects. Doctoral thesis. Lund University, Department of Cell and Organism Biology.

Frederiksen R, Wcislo WT & Warrant EJ (2008) Visual reliability and information rate in the retina of a nocturnal bee. Curr Biol 18: 349-353.

French AS (1979) The effect of light adaptation on the dynamic properties of phototransduction in the fly, Phormia regina. Biol Cybern 32: 115-123.

French AS (1980a) Coherence improvement in white noise analysis by the use of a repeated random sequence generator. IEEE Trans Biomed Eng 27: 51-53.

French AS (1980b) The linear dynamic properties of phototransduction in the fly compound eye. J Physiol 308: 385-401.

French AS (1980c) Phototransduction in the fly compound eye exhibits temporal resonances and a pure time delay. Nature 283: 200-202.

French AS (1992) Mechanotransduction. Annu Rev Physiol 54: 135-152. French AS & Järvilehto M (1978) The dynamic behaviour of photoreceptor cells

in the fly in response to random (white noise) stimulation at a range of temperatures. J Physiol 274: 311-322.

French AS, Snyder AW & Stavenga DG (1977) Image degradation by an irregular retinal mosaic. Biol Cybern 27: 229-233.

Fuortes MG & Yeandle S (1964) Probability of occurrence of discrete potential waves in the eye of Limulus. J Gen Physiol 47: 443-463.

85

Füller H, Eckert M & Blechschmidt K (1989) Distribution of GABA-like immunoreactive neurons in the optic lobes of Periplaneta americana. Cell Tissue Res 255: 225-233.

Gold GH (1999) Controversial issues in vertebrate olfactory transduction. Annu Rev Physiol 61: 857-871.

Goldsmith TH & Ruck PR (1958) The spectral sensitivities of the dorsal ocelli of cockroaches and honeybees: an electrophysiological study. J Gen Physiol 41: 1171-1185.

Goldstein RS & Camhi JM (1988) Modulation of activity in sensory neurons and wind-sensitive interneurons by cercal displacement in the cockroach. J Comp Physiol A 163: 479-487.

Grandcolas P (1996) The phylogeny of cockroach families: a cladistic appraisal of morpho-anatomical data. Can J Zool 74: 508-527.

Greiner B (2006) Visual adaptations in the night-active wasp Apoica pallens. J Comp Neurol 495: 255-262.

Greiner B, Ribi WA & Warrant EJ (2004a) Retinal and optical adaptations for nocturnal vision in the halictid bee Megalopta genalis. Cell Tissue Res 316: 377-390.

Greiner B, Ribi WA & Warrant EJ (2005) A neural network to improve dim-light vision? Dendritic fields of first-order interneurons in the nocturnal bee Megalopta genalis. Cell Tissue Res 322: 313-320.

Greiner B, Ribi WA, Wcislo WT & Warrant EJ (2004b) Neural organisation in the first optic ganglion of the nocturnal bee Megalopta genalis. Cell Tissue Res 318: 429-437.

Guthrie DM & Tindall AR (1968) The biology of the cockroach. Edward Arnold (Publishers) Ltd., London.

Halloy J, Sempo G, Caprari G, Rivault C, Asadpour M, Tache F, Said I, Durier V, Canonge S, Ame JM, Detrain C, Correll N, Martinoli A, Mondada F, Siegwart R & Deneubourg JL (2007) Social integration of robots into groups of cockroaches to control self-organized choices. Science 318: 1155-1158.

Hardie RC (1986) The photoreceptor array of the dipteran retina. Trends Neurosci 9: 419-423.

Hardie RC (2001) Phototransduction in Drosophila melanogaster. J Exp Biol 204: 3403-3409.

Hardie RC (2006) Phototransduction in invertebrate photoreceptors. In: E. J. Warrant and D.-E. Nilsson (eds) Invertebrate vision. Cambridge, Gambridge University Press, p 43–82.

Hardie RC & Raghu P (2001) Visual transduction in Drosophila. Nature 413: 186-193.

Harris FJ (1978) On the use of the windows for harmonic analysis with the discrete Fourier transform. Proceedings of the IEEE 66: 51-84.

86

Heimonen K (1990) Torakan (Periplaneta americana) verkkosilmän näkösolujen toiminallinen identifiointi ja elektrofysiologiset ominaisuudet (engl. Functional indentification and electrophysiological properties of the photoreceptors in the compound eye of the cockroach). Master thesis. University of Oulu, Department of Zoology, Section of Physiology.

Heimonen K (1999) Valoinformaation käsittely hämärään sopeutuneen hyönteisen (Periplaneta americana) näköjärjestelmässä (engl. Processing of light information in the visual system of an insect adapted to dim environment). Licentiate thesis. University of Oulu, Department of Biology, Section of Animal Physiology.

Heimonen K, Mäkelä S-M & Weckström M (2001) Passive cable model of cockroach photoreceptors: impedance functions and estimation of the site of action potential generation. International Conference on Invertebrate Vision, Bäckaskog Castle, Sweden, Lund University: 162.

Henderson SR, Reuss H & Hardie RC (2000) Single photon responses in Drosophila photoreceptors and their regulation by Ca2+. J Physiol 524: 179-194.

Herness MS & Gilbertson TA (1999) Cellular mechanisms of taste transduction. Annu Rev Physiol 61: 873-900.

Hess RF, Sharpe LT & Nordby K (1990) Night vision: basic, clinical and applied aspects. Cambridge University Press, Cambridge, New York, Port Chester, Melbourne, Sydney.

Hille B (2001) Ion channels of excitable membranes. Sinauer, Sunderland. Horridge GA & Barnard PB (1965) Movement of palisade in locust retinula cells

when illuminated. Q J Microsc Sci 106: 131-135. Howard J, Blakeslee B & Laughlin SB (1987) The intracellular pupil mechanism

and photoreceptor signal: noise ratios in the fly Lucilia cuprina. Proc R Soc Lond B Biol Sci 231: 415-435.

Howard J, Dubs A & Payne R (1984) The dynamics of phototransduction in insects: A comparative study. J Comp Physiol A 154: 707-718.

Huber I, Masler EP & Rao BR (eds) (1990a). Cockroaches as models for neurobiology: applications in biomedical research. Vol. I. Boca Raton, Florida, CRC Press, Inc.

Huber I, Masler EP & Rao BR (eds) (1990b). Cockroaches as models for neurobiology: applications in biomedical research. Vol. II. Boca Raton, Florida, CRC Press, Inc.

Johnston D & Wu SM-s (1995) Foundations of cellular neurophysiology. MIT Press, Cambridge, Massachusetts.

Jokela-Määttä M, Pahlberg J, Lindström M, Zak PP, Porter M, Ostrovsky MA, Cronin TW & Donner K (2005) Visual pigment absorbance and spectral sensitivity of the Mysis relicta species group (Crustacea, Mysida) in different light environments. J Comp Physiol A 191: 1087-1097.

87

Juusola M (1993a) Contrast coding in dipteran photoreceptors: The role of phototransduction and the cell membrane. Acta Univ Oul D 272. Dortoral thesis. University of Oulu, Department of Physiology.

Juusola M (1993b) Linear and nonlinear contrast coding in light-adapted blowfly photoreceptors. J Comp Physiol A 172: 511-521.

Juusola M, French AS, Uusitalo RO & Weckstrom M (1996) Information processing by graded-potential transmission through tonically active synapses. Trends Neurosci 19(7): 292-7.

Juusola M & Hardie RC (2001a) Light adaptation in Drosophila photoreceptors: I. Response dynamics and signaling efficiency at 25 degrees C. J Gen Physiol 117: 3-25.

Juusola M & Hardie RC (2001b) Light adaptation in Drosophila photoreceptors: II. Rising temperature increases the bandwidth of reliable signaling. J Gen Physiol 117: 27-42.

Juusola M, Kouvalainen E, Järvilehto M & Weckström M (1994) Contrast gain, signal-to-noise ratio, and linearity in light-adapted blowfly photoreceptors. J Gen Physiol 104: 593-621.

Juusola M, Uusitalo RO & Weckström M (1995) Transfer of graded potentials at the photoreceptor-interneuron synapse. J Gen Physiol 105: 117-148.

Juusola M & Weckström M (1993) Band-pass filtering by voltage-dependent membrane in an insect photoreceptor. Neurosci Lett 154: 84-88.

Järvilehto M (1978) Light information transmission and processing in the visual system of the fly. Acta Univ Oul D 40. Doctoral thesis. University of Oulu, Department of Physilogy and Department of Zoology.

Järvilehto M (1985) The eye: vision and perception. In: G. A. Kerkut and L. I. Gilbert (eds) Comprehensive insect physiology, biochemistry and pharmacology. Oxford, Pergamon Press Vol. 6. Nervous system: sensory, p 356-427.

Järvilehto M & Zettler F (1973) Electrophysiological-histological studies on some functional properties of visual cells and second order neurons of an insect retina. Z Zellforsch 136: 291-306.

Kambhampati S (1995) A phylogeny of cockroaches and related insects based on DNA-sequence of mitochondrial ribosomal-RNA genes. Proc Nat Acad Sci 92: 2017-2020.

Kandel ER, Schwartz JH & Jessell TM (eds) (2000). Principles of neural science. New York, McGraw-Hill.

Kauranen M & Weckström M (2004) K+ channels and their modulation by 5-HT in Drosophila photoreceptors: a modelling study. Ann Biomed Eng 32: 1580-1595.

Kelber A (2006) Invertebrate colour vision. In: E. J. Warrant and D.-E. Nilsson (eds) Invertebrate vision. Cambridge, Gambridge University Press, p 250–290.

88

Kelber A, Balkenius A & Warrant EJ (2002) Scotopic colour vision in nocturnal hawkmoths. Nature 419: 922-925.

Kelly K & Mote MI (1990a) Avoidance of monochromatic light by the cockroach Periplaneta americana. J Insect Physiol 36: 287-291.

Kelly KM & Mote MI (1990b) Electrophysiology and anatomy of medulla interneurons in the optic lobe of the cockroach, Periplaneta americana. J Comp Physiol A 167: 745-756.

Kettunen P (1994) Torakan (Periplaneta americana) optomotorinen vaste hämärään sopeutuneen näköjärjestelmän mallina (engl. The optomotor response of the cockroach as a model of a visual system adapted to dim light). Master thesis. University of Oulu, Department of Zoology, Section of Animal Physiology.

Kirschfeld K (1971) Aufnahme und Verarbeitung optischer Daten im Komplexauge von Insekten. Naturwissenschaften 58: 201-209.

Kirschfeld K & Franceschini N (1969) Ein Mechanismus zur Steuerung des Lichtflusses in den Rhabdomeren des Komplexauges von Musca. Kybernetik 6: 13-22.

Koch C (1999) Biophysics of computation: information processing in single neurons. Oxford University Press, New York.

Kouvalainen E, Weckström M & Juusola M (1994) A method for determining photoreceptor signal-to-noise ratio in the time and frequency domains with a pseudorandom stimulus. Vis Neurosci 11: 1221-1225.

Krause S, Krause Y, Immonen E-V & Weckström M (2008a) Visual transduction in cockroach photoreceptors is mediated by TRPL but not TRP channels. The Second International Conference on Invertebrate Vision, Bäckaskog Castle, Sweden, Lund University: 36.

Krause Y, Krause S, Huang J, Liu CH, Hardie RC & Weckström M (2008b) Light-dependent modulation of Shab channels via phosphoinositide depletion in Drosophila photoreceptors. Neuron 59: 596-607.

Kuster JE & French AS (1985) Changes in the dynamic properties of the locust photoreceptors at three levels of light adaptation. Biol Cybern 52: 333-337.

Kwon HW, Lent DD & Strausfeld NJ (2004) Spatial learning in the restrained American cockroach Periplaneta americana. J Exp Biol 207: 377-383.

Land MF (1981) Optics and vision in invertebrates. In: H. Autrum (ed) Handbook of Sensory Physiology VII/6B, p 471–592.

Land MF (1985) The eye: optics. In: G. A. Kerkut and L. I. Gilbert (eds) Comprehensive insect physiology, biochemistry and pharmacology. Oxford, Pergamon Press Vol. 6, p 225-275.

Land MF (1997) Visual acuity in insects. Annu Rev Entomol 42: 147-177. Laughlin SB (1981) Neural principles in the peripheral visual systems of

invertebrates. In: H. Autrum (ed) Handbook of sensory physiology. Berlin, Heidelberg, New York, Springer-Verlag Vol. VII/6B, p 133-280.

89

Laughlin SB (1987) Form and function in retinal processing. Trends Neurosci 10: 478-483.

Laughlin SB (1989) The role of sensory adaptation in the retina. J Exp Biol 146: 39-62.

Laughlin SB (ed) (1990). Invertebrate vision at low luminances. Night vision. Cambridge, Cambridge University Press.

Laughlin SB (1996) Matched filtering by a photoreceptor membrane. Vision Res 36: 1529-1541.

Laughlin SB (2001) Energy as a constraint on the coding and processing of sensory information. Curr Opin Neurobiol 11: 475-480.

Laughlin SB, de Ruyter van Steveninck RR & Anderson JC (1998) The metabolic cost of neural information. Nat Neurosci 1: 36-41.

Laughlin SB & Hardie RC (1978) Common strategies for light adaptation in the peripheral visual systems of fly and dragonfly. J Comp Physiol A 128: 319-340.

Laughlin SB & Lillywhite PG (1982) Intrinsic noise in locust photoreceptors. J Physiol 332: 25-45.

Laughlin SB & Weckström M (1993) Fast and slow photoreceptors - a comparative study of the functional diversity of coding and conductances in the Diptera. J Comp Physiol A 172: 593-609.

Lent DD & Kwon HW (2004) Antennal movements reveal associative learning in the American cockroach Periplaneta americana. J Exp Biol 207: 369-375.

Levi R & Camhi JM (2000) Wind direction coding in the cockroach escape response: winner does not take all. J Neurosci 20(10): 3814-21.

Libersat F & Camhi JM (1988) Control of cercal position during flight in the cockroach: a mechanism for regulating sensory feedback. J Exp Biol 136: 483-488.

Lillywhite PG (1977) Single photon signals and transduction in insect eye. J Comp Physiol A 122: 189-200.

Lillywhite PG & Laughlin SB (1979) Transducer noise in a photoreceptor. Nature 277: 569-572.

Lipetz LE (1971) The relation of physiological and psychological aspects of sensory intensity. In: W. R. Loewenstein (ed) Handbook of sensory physiology. Berlin, Heidelberg, New York, Springer-Verlag Vol. I. Principles of receptor physiology, p 191-225.

Lipton GR & Sutherland DJ (1970) Activity rhythms in the American cockroach, Periplaneta americana. J Insect Physiol 16: 1555-1566.

Lumpkin EA & Caterina MJ (2007) Mechanisms of sensory transduction in the skin. Nature 445: 858-865.

Marmarelis PZ & Marmarelis VZ (1978) Analysis of physiological systems: The white noise approach. Plenum Press, New York.

90

Matic T & Laughlin SB (1981) Changes in the intensity-response function of an insect's photoreceptors due to light adaptation. J Comp Physiol A 145: 169-177.

Menzel R (1979) Spectral sensitivity and color vision in invertebrates. In: H. Autrum (ed) Handbook of Sensory Physiology. Berlin, Springer-Verlag Vol. VII/6A, p 503-580.

Menzel R & Backhaus W (1989) Color vision in honey bees: phenomena and physiological mechanisms. In: D. G. Stavenga and R. C. Hardie (eds) Facets of vision. Berlin, Heidelberg, New York, Springer-Verlag, p 281-297.

Meyer NE, Joel-Almagor T, Frechter S, Minke B & Huber A (2006) Subcellular translocation of the eGFP-tagged TRPL channel in Drosophila photoreceptors requires activation of the phototransduction cascade. J Cell Sci 119: 2592-2603.

Minke B & Parnas M (2006) Insights on TRP channels from in vivo studies in Drosophila. Annu Rev Physiol 68: 649-684.

Mishra M & Meyer-Rochow VB (2008) Fine structural description of the compound eye of the Madagascar 'hissing cockroach' Gromphadorhina portentosa (Dictyoptera: Blaberidae). Insect Science 15: 179-192.

Mizunami M (1990) Nonlinear signal transmission between second- and third-order neurons of cockroach ocelli. J Gen Physiol 95: 297-317.

Mizunami M (1994) Information processing in the insect ocellar system: comparative approaches to the evolution of visual processing and neural circuits. Advanges in Insect Physiology 25: 151-265.

Mizunami M (1995a) Functional diversity of neural organization in insect ocellar systems. Vision Res 35: 443-452.

Mizunami M (1995b) Morphology of higher-order ocellar interneurons in the cockroach brain. J Comp Neurol 362: 293-304.

Mizunami M (1995c) Neural organization of ocellar pathways in the cockroach brain. J Comp Neurol 352: 458-468.

Mizunami M (1996) Gain control of synaptic transfer from second- to third-order neurons of cockroach ocelli. J Gen Physiol 107: 121-131.

Mizunami M & Tateda H (1986) Classification of Ocellar Interneurones in the Cockroach Brain. J Exp Biol 125: 57-70.

Mizunami M & Tateda H (1988a) Dynamic relationship between the slow potential and spikes in cockroach ocellar neurons. J Gen Physiol 91: 703-723.

Mizunami M & Tateda H (1988b) Synaptic transmission between second- and third-order neurones of cockroach ocelli. J Exp Biol 140: 557-561.

Mizunami M, Tateda H & Naka K (1986) Dynamics of cockroach ocellar neurons. J Gen Physiol 88: 275-292.

91

Mizunami M, Weibrecht JM & Strausfeld NJ (1998) Mushroom bodies of the cockroach: their participation in place memory. J Comp Neurol 402: 520-537.

Mizunami M, Yamashita S & Tateda H (1982) Intracellular stainings of the large ocellar second order neurons in the cockroach. J Comp Physiol A 149: 215-219.

Mote MI (1990) Structure and function of the visual systems of the American cockroach. In: I. Huber, E. P. Masler and B. R. Rao (eds) Cockroaches as models for neurobiology: applications in biomedical research. Boca Raton, Florida, CRC Press, Inc. Vol. II, p 203-224.

Mote MI & Goldsmith TH (1970) Spectral sensitivities of color receptors in the compound eye of the cockroach Periplaneta. J Exp Zool 173: 137-145.

Mote MI & Goldsmith TH (1971) Compound eyes: localization of two color receptors in the same ommatidium. Science 171: 1254-1255.

Mote MI, Kumar VSN & Black KR (1981) ‘On’ type interneurons in the optic lobe of Periplaneta americana. II. Receptive fields and response latencies. J Comp Physiol A 141: 403-415.

Mote MI & Rubin LJ (1981) ‘On’ type interneurons in the optic lobe of Periplaneta americana. I. Spectral characteristics of response. J Comp Physiol A 141: 395-401.

Naka KI & Rushton WA (1966) S-potentials from colour units in the retina of fish (Cyprinidae). J Physiol 185: 536-555.

Nilsson DE (1989) Optics and evolution of the compound eye. In: D. G. Stavenga and R. C. Hardie (eds) Facets of Vision. Berlin, Heidelberg, Springer-Verlag, p 30-73.

Nilsson DE (1990) From cornea to retinal image in invertebrate eyes. Trends Neurosci 13: 55-64.

Nilsson DE & Kelber A (2007) A functional analysis of compound eye evolution. Arthropod Struct Dev 36: 373-385.

Nishino H & Mizunami M (1998) Giant input neurons of the mushroom body: intracellular recording and staining in the cockroach. Neurosci Lett 246: 57-60.

Niven JE, Anderson JC & Laughlin SB (2007) Fly photoreceptors demonstrate energy-information trade-offs in neural coding. PLoS Biol 5: e116.

Niven JE & Laughlin SB (2008) Energy limitation as a selective pressure on the evolution of sensory systems. J Exp Biol 211: 1792-1804.

Niven JE, Vähäsöyrinki M, Kauranen M, Hardie RC, Juusola M & Weckström M (2003) The contribution of Shaker K+ channels to the information capacity of Drosophila photoreceptors. Nature 421: 630-634.

Nowel MS (1981) Postembryonic growth of the compound eye of the cockroach. J Embryol exp Morph 62: 259-275.

Ohyama T & Toh Y (1986) Multimodality of ocellar interneurons of the American cockroach. J Exp Biol 125: 405-409.

92

Okada J & Toh Y (1998) Shade response in the escape behavior of the cockroach, Periplaneta americana. Zool Sci 15: 831-835.

Okada J & Toh Y (2006) Active tactile sensing for localization of objects by the cockroach antenna. J Comp Physiol A 192: 715-726.

Page TL (1990) Circadian organisation in the cockroach. In: I. Huber, E. P. Masler and B. R. Rao (eds) Cockroaches as models for neurobiology: applications in biomedical research. Boca Raton, Florida, CRC Press, Inc. Vol. II, p 225-245.

Pahlberg J, Lindström M, Ala-Laurila P, Fyhrquist-Vanni N, Koskelainen A & Donner K (2005) The photoactivation energy of the visual pigment in two spectrally different populations of Mysis relicta (Crustacea, Mysida). J Comp Physiol A 191: 837-844.

Parker A (2004) In the blink of an eye: how vision kick-started the big bang of evolution. Simon & Schuster UK Ldt, London.

Pelli DG (1990) The quantum efficiency of vision. Vision: Coding and Efficiency: 3–24.

Pick B & Buchner E (1979) Visual movement detection under light- and dark-adaptation in the fly, Musca domestica. J Comp Physiol A 134: 45-54.

Ribi WA (1977) Fine structure of the first optic ganglion (lamina) of the cockroach, Periplaneta americana. Tissue Cell 9: 57-72.

Ribi WA (1979) The first optic ganglion of the bee. III. Regional comparison of the morphology of photoreceptor-cell axons. Cell Tissue Res 200: 345-357.

Ribi WA (1981) The first optic ganglion of the bee. IV. Synaptic fine structure and connectivity patterns of receptor cell axons and first order interneurones. Cell Tissue Res 215: 443-464.

Riemay H (1984) Verhaltensphysiologische Untersuchungen zur Einwirkung des Lichtes auf den Fluchtlauf der Schabe Periplaneta americana (L.). Zool Jb Physiol 88: 491-497.

Rivault C (1983) Role of photoreceptors in the synchronization of the circadian locomotor activity rhythms with the LD cycle in cockroaches (Periplaneta americana). J interdiscipl Cycle Res 14: 21-31.

Roberts SK (1965) Photoreception and entrainment of cockroach activity rhythms. Science 148: 958-959.

Roebroek JG & Stavenga DG (1990a) On the effective optical density of the pupil mechanism in fly photoreceptors. Vision Res 30: 1235-1242.

Roebroek JGH & Stavenga DG (1990b) Insect pupil mechanisms IV. Spectral characteristics and light intensity dependence in the blowfly Calliphora erythrocephala. J Comp Physiol A 166: 537-543.

Rushton WAH (1972) Visual pigments in man. In: H. J. A. Dartnall (ed) Handbook of Sensory Physiology. New York, Springer-Verlag Vol. VII/1. Photochemistry of Vision, p 364-394.

93

Salmela I, Heimonen K & Weckström M (2004) Hodgkin-Huxley type compartmental modelling of cockroach photoreceptors with graded and spike responses. J Physiol 561P: PC14.

Scholes JH (1964) Discrete subthreshold potentials from the dimly lit insect eye. Nature 202: 572-573.

Seelinger G (1990) Chemoreception. In: I. Huber, E. P. Masler and B. R. Rao (eds) Cockroaches as models for neurobiology: applications in biomedical research. Boca Raton, Florida, CRC Press, Inc. Vol. II, p 296-284.

Shannon CE & Weaver W (1949) The mathematical theory of communication. University of Illinois Press, Urbana, Chicago.

Shaw SR (1984) Early visual processing in insects. J Exp Biol 112: 225-251. Shaw SR (1989) The retina-lamina pathway in insects, particularly Diptera,

viewed from an evolutionary perspective. In: D. G. Stavenga and R. C. Hardie (eds) Facets of Vision. Berlin, Heidelberg, Springer-Verlag, p 186-212.

Shaw SR (1990) The photoreceptor axon projection and its evolution in the neural superposition eyes of some primitive brachyceran Diptera. Brain Behav Evol 35: 107-125.

Smola U (1976) Voltage noise in insect visual cells. In: F. Zettler and R. Weiler (eds) Neural principles in vision. Berlin, Heidelberg, New York, Springer-Verlag, p 194-213.

Snyder AW (1975) Photoreceptor optics - theoretical principles. In: A. W. Snyder and R. Menzel (eds) Photoreceptor optics. Berlin, Heidelberg, New York, Springer-Verlag, p 38-55.

Snyder AW & Horridge GA (1972) The optical function of changes in the medium surrounding the cockroach rhabdom. J Comp Physiol A 81: 1-8.

Snyder AW, Menzel R & Laughlin SB (1973) Structure and function of the fused rhabdom. J Comp Physiol A 87: 99-135.

Stark RJ & Mote MI (1981) Postembryonic development of the visual system of Periplaneta americana. I. Patterns of growth and differentiation. J Embryol Exp Morphol 66: 235-255.

Stavenga DG (1979) Pseudopupils of compound eyes. In: H. Autrum (ed) Handbook of sensory physiology. Berlin, Heidelberg, New York, Springer-Verlag Vol VII/6A, p 357-439.

Stavenga DG (2003a) Angular and spectral sensitivity of fly photoreceptors. I. Integrated facet lens and rhabdomere optics. J Comp Physiol A 189: 1-17.

Stavenga DG (2003b) Angular and spectral sensitivity of fly photoreceptors. II. Dependence on facet lens F-number and rhabdomere type in Drosophila. J Comp Physiol A 189: 189-202.

Stavenga DG (2004) Angular and spectral sensitivity of fly photoreceptors. III. Dependence on the pupil mechanism in the blowfly Calliphora. J Comp Physiol A 190: 115-129.

94

Stavenga DG & Hardie RC (eds) (1989). Facets of Vision. Berlin, Heidelberg, Springer-Verlag.

Stavenga DG & Schwemer J (1984) Visual pigments of invertebrates. In: M. A. Ali (ed) Photoreception and vision in invertebrates. New York, London, Plenum Press, p 11-61.

Stewart WW (1978) Functional connections between cells as revealed by dye-coupling with a highly fluorescent naphthalimide tracer. Cell 14: 741-759.

Stewart WW (1981) Lucifer dyes - highly fluorescent dyes for biological tracing. Nature 292: 17-21.

Strausfeld N, Douglass J, Campbell H & Higgins C (2006) Parallel processing in the optic lobes of flies and the occurence of motion computing circuits. In: E. J. Warrant and D.-E. Nilsson (eds) Invertebrate vision. Cambridge, Gambridge University Press, p 349–398.

Toh Y & Yokohari F (1988) Postembryonic development of the dorsal ocellus of the American cockroach. J Comp Neurol 269: 157-167.

Trujillo-Cenoz O & Melamed J (1971) Spatial distribution of photoreceptor cells in the ommatidia of Periplaneta americana. J Ultrastruct Res 34: 397-400.

Uusitalo R (1995) Visual signal processing mechanisms in the dipteran compound eye. The role of the first visual synapse and the first visual interneurons. Acta Univ Oul D 338. Doctoral thesis. University of Oulu, Department of Physilogy.

Vallet AM & Coles JA (1991) A method for estimating the minimum visual stimulus that evokes a behavioural response in the drone, Apis mellifera male. Vision Res 31: 1453-1455.

Vallet AM & Coles JA (1993) Is the membrane voltage amplifier of drone photoreceptors useful at physiological intensities? J Comp Physiol A 173: 163-168.

Vallet AM, Coles JA, Eilbeck JC & Scott AC (1992) Membrane conductances involved in amplification of small signals by sodium channels in photoreceptors of drone honey bee. J Physiol 456: 303-324.

Walther JB (1958a) Changes induced in spectral sensitivity and form of retinal action potential of the cockroach eye by selective adaptation. J Insect Physiol 2: 142-151.

Walther JB (1958b) Untersuchungen am Belichtungspotential des Komplexauges von Periplaneta mit farbigen Reizen und selektiver Adaptation. Biol Zbl 77: 63-104.

van Hateren JH (1986) An efficient algorithm for cable theory, applied to blowfly photoreceptor cells and LMC's. Biol Cybern 54: 301-311.

van Hateren JH (1992) A theory of maximizing sensory information. Biol Cybern 68: 23-29.

95

van Hateren JH & Laughlin SB (1990) Membrane parameters, signal transmission, and the design of a graded potential neuron. J Comp Physiol A 166: 437-448.

Warrant E (2004) Vision in the dimmest habitats on earth. J Comp Physiol A 190: 765-789.

Warrant EJ (1999) Seeing better at night: life style, eye design and the optimum strategy of spatial and temporal summation. Vision Res 39: 1611-1630.

Warrant EJ (2006) Invertebrate vision in dim light. In: E. J. Warrant and D.-E. Nilsson (eds) Invertebrate vision. Cambridge, Gambridge University Press, p 83–126.

Warrant EJ (2008) Seeing in the dark: vision and visual behaviour in nocturnal bees and wasps. J Exp Biol 211: 1737-1746.

Warrant EJ, Kelber A, Gislen A, Greiner B, Ribi W & Wcislo WT (2004) Nocturnal vision and landmark orientation in a tropical halictid bee. Curr Biol 14: 1309-1318.

Warrant EJ & McIntyre PD (1992) The trade-off between resolution and sensitivity in compound eye. In: R. B. Pinter and B. Nabet (eds) Nonlinear vision: determination of neural receptive fields, function, and networks. Boca Raton, Ann Arbor, London, Tokyo, CRC Press, p 391-421.

Warrant EJ & Nilsson D-E (eds) (2006). Invertebrate vision. Cambridge, Gambridge University Press.

Weckström M (1987) Temperature dependence, dark adaptation and the dynamics of phototransduction in dipteran photoreceptors. Acta Univ Oul D 163 thesis. University of Oulu, Department of Physilogy.

Weckström M, Hardie RC & Laughlin SB (1991) Voltage-activated potassium channels in blowfly photoreceptors and their role in light adaptation. J Physiol 440: 635-657.

Weckström M, Juusola M, Uusitalo RO & French AS (1995) Fast-acting compressive and facilitatory nonlinearities in light-adapted fly photoreceptors. Ann Biomed Eng 23: 70-77.

Weckström M, Kouvalainen E & Juusola M (1992) Measurement of cell impedance in frequency domain using discontinuous current clamp and white-noise-modulated current injection. Pflugers Arch 421: 469-472.

Weckström M, Kouvalainen E & Järvilehto M (1988) Non-linearities in response properties of insect visual cells: an analysis in time and frequency domain. Acta Physiol Scand 132: 103-113.

Weckström M & Laughlin SB (1995) Visual ecology and voltage-gated ion channels in insect photoreceptors. Trends Neurosci 18: 17-21.

Willis MA & Avondet JL (2005) Odor-modulated orientation in walking male cockroaches Periplaneta americana, and the effects of odor plumes of different structure. J Exp Biol 208: 721-735.

96

Willis MA, Avondet JL & Finnell AS (2008) Effects of altering flow and odor information on plume tracking behavior in walking cockroaches, Periplaneta americana (L.). J Exp Biol 211: 2317-2326.

Wolken JJ (1971) Invertebrate photoreceptors. Academic Press, New York, London.

Vollrath MA, Kwan KY & Corey DP (2007) The micromachinery of mechanotransduction in hair cells. Annu Rev Neurosci 30: 339-365.

Wong F & Knight BW (1980) Adapting-bump model for eccentric cells of Limulus. J Gen Physiol 76: 539-557.

Wong F, Knight BW & Dodge FA (1980) Dispersion of latencies in photoreceptors of Limulus and the adapting-bump model. J Gen Physiol 76: 517-537.

Wong F, Knight BW & Dodge FA (1982) Adapting bump model for ventral photoreceptors of Limulus. J Gen Physiol 79: 1089-1113.

Voutilainen S (1996) Torakan (Periplaneta americana) optomotorinen kääntymisvaste: taajuusfunktio ja adaptaatio (engl. The optomotor turning response of the cockroach: frequency function and adaptation). Master thesis. University of Oulu, Department of Zoology, Section of Animal Physiology.

Wu C-F & Pak L (1978) Light induced voltage noise in the photoreceptor of Drosophila melanogaster. J Gen Physiol 71: 249-268.

Vähäsöyrinki M (2004) Voltage-gated K+ channels in Drosophila photoreceptors. Biophysical study of neural coding. Acta Univ Oul A 426. Doctoral thesis. University of Oulu, Department of Physical Sciences, Biophysics.

Vähäsöyrinki M, Niven JE, Hardie RC, Weckström M & Juusola M (2006) Robustness of neural coding in Drosophila photoreceptors in the absence of slow delayed rectifier K+ channels. J Neurosci 26: 2652-2660.

Ye S & Comer CM (1996) Correspondence of escape-turning behavior with activity of descending mechanosensory interneurons in the cockroach, Periplaneta americana. J Neurosci 16: 5844-5853.

Ye S, Leung V, Khan A, Baba Y & Comer CM (2003) The antennal system and cockroach evasive behavior. I. Roles for visual and mechanosensory cues in the response. J Comp Physiol A 189: 89-96.

Zettler F & Järvilehto M (1973) Active and passive axonal propagation of non-spike signals in the retina of Calliphora. J Comp Physiol A 85: 89-104.

Zill SN (1990) Mechanoreceptors: exteroceptors and proprioceptors. In: I. Huber, E. P. Masler and B. R. Rao (eds) Cockroaches as models for neurobiology: applications in biomedical research. Boca Raton, Florida, CRC Press, Inc. Vol. II, p 247-268.

Zill SN & Seyfarth E-A (1996) Exoskeletal sensors for walking. Sci Am 275: 70-74.

97

Original articles

The original articles have been reprinted with the permissions of

The American Physiological Society (www.the-aps.org);

Journal of Neurophysiology (jn.physiology.org)

and Society for Neuroscience (www.sfn.org);

Journal of Neuroscience (www.jneurosci.org)

processing of visual information in dim light. functional variability...

Documents