haggard & chambon 2012 sense of agency

7/29/2019 Haggard & Chambon 2012 Sense of Agency

1/3

Current BiologyVol 22 No 10R390

agency: that is, what it eels like to be

in control o the image viewed on the

screen, as opposed to out o control.

What brain computations underlie

sense o agency? The experiments

described above suggest the brain

computes agency by predicting the

consequences o current actions,

and comparing these predictions

to actual outcomes (Figure 1A). In

the example above, knowing how

one moves the joystick allows one

to predict the visual eedback. I

the video eedback matches the

prediction, then it ollows that I did

that. I there is no match, then the

visual eedback must have another

cause. This account is based on the

computations used to monitor and

adjust motor commands in a popular

computational ramework or action

control, leading to the suggestionthat sense o agency is simply the

normal operation o the goaldirected

action system (Farreret al. 2008). In

this ramework, agency is computed

by comparing a prospective signal

(the intention or command to achieve

the goal) and a retrospective signal

(the actual eedback or outcome o

action). An internal orward model

uses a copy o the current motor

command to predict the eedback

that the action will produce. I the

comparison between predicted and

actual eedback generates no error,then I did that.

Brain imaging studies o such

tasks have not ound any clear

positive correlate o agency,

but routinely show activation o

the angular gyrus in the parietal

cortex in situations o nonagency.

Angular gyrus activation increases

proportionately as the subjective

sense o agency decreases. This

brain area may thereore either house

the comparator, or receive the error

signals transmitted by a comparator

located somewhere else in the brain.

These studies suggest a strong

link between sense o agency and

errormonitoring in the motor control

system, but they leave two important

questions unanswered. First, why do

we have a positive sense o agency,

in addition to the negative sense

o nonagency that occurs when

the outcome o an action is not as

predicted? Second, what experience

is there prior to receiving delayed

eedback about action outcomes:

could an experience similar to agency

Sense of agency

Patrick Haggard1

and Valerian Chambon1,2

What is the sense o agency? The

term sense o agency reers to the

experience o controlling ones own

actions, and, through them, events in

the outside world. Most o us have the

eeling that we are in control o what

we are doing most o the time: this is

the normal sense o agency.

Why is it important? Some many

mental illnesses involve abnormalities

o agency: psychotic patients

sometimes report that their actions

are not their own, but are imposed

on them by some other agent, while

depressed patients oten experiencehelplessness and loss o agency.

Sense o agency also plays an

important role in society generally,

because it is central to the idea o

responsibility or our own actions. In

many countries, the law requires that a

person be aware o the consequences

o their actions, i they are to be ound

guilty o a crime.

How can sense o agency be studied

scientically? Despite its importance,

sense o agency has proved dicult

to study scientically. Loss o agency,or example when one is driving a

car that suddenly malunctions, is a

salient, and dramatic experience. But

the normal fow o action and control

orms a continuous, thin background

to consciousness, rather than a

specic, identiable experience.

Psychology and neuroscience

have struggled to measure sense

o agency. Most studies have used

explicit agency attribution tasks,

in which a person judges whether

they did or did not cause a specic

event. One experimental paradigm,

or example, involves participants

moving a joystick, while watching

video eedback. They judge whether

they caused the joystick movement

shown on the video or not, equivalent

to evaluating the truth or alsity o

the proposition I did that. These

studies have shown that spatial and

temporal contiguity between ones

own and the observed movement

are the major cues or selattributing

agency; however, they say little about

thephenomenologyor experience o

in a nearly complete overthrow o

the seedless vascular plants, which

once dominated the planet. The

modernday descendants o these

giants, still dependent on water or

hal o their lie cycle, are almost

exclusively conned to swampy or

tropical habitats. Equisetum, with

their highly prolierative and resilient

rhizomes, are more versatile than

many species and can be ound in

various environments.

How do humans use Equisetum?

In many cultures, the rough stems

o horsetails are dried and used to

scour and polish metal pots and pans,

earning them the name scouring

rush in some parts o the world.

In addition, traditional Japanese

woodworkers use dried Equisetum as

a type o extrane sandpaper. Since

Equisetum species can be ound allover the globe (except Australasia),

they have made their way into the

traditional medicine practices o

various cultures, dating back to

ancient Greek and Roman times.

The silica in the stems o horsetails

is thought to aid in osteoporosis

prevention, while extracts rom

horsetails are used or many dierent

reasons, including as a diuretic, and

as a source o natural antioxidants.

Very ew studies, though, have been

perormed to weigh the merit o

these alternative medical uses or theplant. Interestingly, livestock have

been known to become ill and even

die ater eating large quantities o

Equisetum or this reason, and

because most commercial pesticides

kill only seedbearing plants,

horsetails are considered to be tricky

pests in some areas.

Where can I nd out more?Beerling, D. (2007). The Emerald Planet: How

Plants Changed Earths History (Oxord: OxordUniversity Press).

Du, R.J., and Nickrent, D.L. (1999). Phylogeneticrelationships o land plants using mitochondrial

smallsubunit rDNA sequences. Am. J. Bot. 86,372386.

Raven, P.H., Evert, R.F., and Eichhorn, S.E. (1999).Biology o Plants (New York: W.H. Freeman andcompany).

Renzaglia, K.S., Du, R.J., Nickrent, D.L.,and Garbary, D.J. (2000). Vegetative andreproductive innovations o early land plants:implications or a unied phylogeny. Phil.Trans. R. Soc. Lon.365, 769793.

http:www.umm.edualtmedarticleshorsetail000257.htm

http:internationalequisetologicalassociation.yolasite.com

Anne Knowlton is Current Biologys

Assistant Editor.

Email:[email protected]
http://www.umm.edu/altmed/articles/horsetail-000257.htmhttp://www.umm.edu/altmed/articles/horsetail-000257.htmhttp://internationalequisetologicalassociation.yolasite.com/http://internationalequisetologicalassociation.yolasite.com/mailto:[email protected]:[email protected]://internationalequisetologicalassociation.yolasite.com/http://internationalequisetologicalassociation.yolasite.com/http://www.umm.edu/altmed/articles/horsetail-000257.htmhttp://www.umm.edu/altmed/articles/horsetail-000257.htm


2/3

MagazineR391

be present at the time o the action

itsel, or even beore it?

When does the brain compute

the sense o agency? Most recent

studies emphasise theretrospective

component o agency or two reasons.

First, the outcome o actions can be

readily manipulated experimentally,

but there are ew methods or studying

intentions. Second, psychological

experiments have shown that we

are oten mistaken about our own

intentions. For example, in an

experiment where the outcomes

o participants action choices

were surreptitiously changed,

participants reported having an

intention corresponding to the actual

outcome, rather than their original

choice (Johansson et al. 2005). On

one extreme view, sense o agency is

really a retrospective conabulation toexplain our own action choices, rather

than a direct experiential consequence

o neural processing in the brains

motor systems. The predictions

generated by orward models in

the motor system may not reach

conscious awareness.

There are, however, several

reasons or thinking that sense

o agency strongly depends on

prospective processes, and is not just

a retrospective conabulation. First,

the brain prospectively represents the

outcomes o actionsbefore actionoccurs. Prior inormation about

action outcomes increases sense o

agency over those outcomes. When

people are shown subliminal prime

stimuli just prior to an action, primes

corresponding to the consequences

o the impending action lead to a

stronger experience o control over

the outcome than noncorresponding

primes. The primes in these

experiments are not consciously

perceived, but are assumed to boost

the representation o the action

outcome.

Could sense o agency be a

metacognitive experience based on

action selection? Recent studies have

suggested that the brain processes

that select between alternative actions

may provide a second prospective

agency signal, in addition to those

generated by predicting action

outcomes. Specically, people eel a

stronger sense o control when they

choose fuently and easily what to

do (Wenke et al. 2010). This nding

suggests that the sense o agency

depends on volitional processes that

necessarily precede action, and so

cannot be purely retrospective. In

one study (Figure 1B), people were

instructed whether to press a key

with the let or right hand, or else

reely chose which hand to press

with. Subliminal arrow primes just

beore these instructions could

either acilitate or interere with these

choices, by making action selection

more or less fuent. Each action

had the eect o making a colour

appear on the screen. Importantly,

dierent colour eects were used

on trials where the primes were

compatible with the instructed or

reelychosen response, and on trials

where the primes were incompatible.

Importantly, the colour eects did not

depend on the primes themselves,

but on the relation between prime and

subsequent action. Ater several trials,

participants rankordered how much

control they elt over the dierent

colour eects they experienced.

They reported stronger sense o

agency over colours that ollowed

compatiblyprimed actions rather

than incompatiblyprimed actions,

irrespective o whether they reely

choose between actions, or were

instructed which action to make.

How might action selection infuence

sense o agency? Most people

recognise situations where they know

exactly what to do, and fuently select

the next appropriate action, like a

welltrained pilot in an emergency.

These situations seem to produce a

strong sense o agency, even beore

the outcome o action is known.

Subliminal priming o actions may

capture this situation experimentally.

Predictedoutcome

Action selection(Inverse model)

Motor commandActionIntention

Predictor

(forwardmodel)

Efference copy

sensory feedback

Outcome

Perception

Sense ofAgency

Actualoutcome

+ -

A

B C

Current Biology

Figure 1. The sense o agency.

(A) Computational motor control rameworks and sense o agency. To achieve a given intention

or goal, an inverse model selects an appropriate motor command. The orward model predicts

the outcome o the motor command, and this is compared to the actual outcome reported by

sensory eedback. A comparator matches prospective signals (red arrows) and retrospective

signals (blue arrow). Zero comparator error corresponds to a sense o agency. A second pro

spective signal generated during action selection directly contributes to the sense o agency

(dashed red arrow). Easy and fuent selection o the appropriate action leads to stronger sense

o agency. (B) Subliminal priming infuences sense o agency over action outcomes. Partici

pants make let or right keypresses in response to target arrows. Brie subliminal arrow primes

precede these targets, and can be either compatible or incompatible with the action indicated

by the target. Keypress responses to the target elicit one o several colour patches, ater a

variable delay. Participants report more control over the colour patch on compatiblyprimed

trials, compared to incompatibly primed trials. (C) A possible circuit or prospective agency.Dorsolateral prerontal cortex selects between alternative possible actions. Confict signals

rom this area are monitored by the angular gyrus, where they are interpreted as a reduced

sense o control. Adapted rom Chambon et al. (2012).


3/3

Current BiologyVol 22 No 10R392

Our recent brain imaging study

(Chambon et al. 2012) ound evidence

that the interaction between the

angular gyrus o the parietal cortex,

and the action selection centres in the

prerontal cortex, plays a key role in

the prospective sense o agency. When

incompatible primes interered with

the decision o which action to make,

the angular gyrus activation at the time

o action selection showed a negative

correlation with sense o agency

over action outcome. This negative

correlation had been observed

beore, and was explained in terms

o predictionoutcome matching. In

contrast, our study ound that angular

gyrus might compute nonagency at

the time o action selection, and not

just retrospectively once outcomes

were known. This study ound no

relation between angular gyrus

activation and sense o control oncompatiblyprimed trials, suggesting

that this area monitors a signal related

to diculty and confict during action

selection, rather than the normal

smooth fow o agency (Figure 1C).

Moreover, the lateral prerontal

cortex, which is responsible or

selecting between dierent possible

actions, showed a negative correlation

with the angular gyrus activation on

incompatiblyprimed only, and no

correlation on compatiblyprimed

trials. These ndings suggest a new

model, in which sense o agencydepends as much on how we choose

what we do, as on actually doing

it. The positive sense o agency, or

eeling in control, could refect the

deault state o the brains action

selection networks, while the eeling

o loss o control could refect angular

gyrus monitoring the confict within

rontal action selection mechanisms.

On this view, the experience o agency

would be a orm o metacognition, or

conscious experience corresponding

to eciency o action selection. In

the computational ramework above,

the planner process, or inverse

model, must contribute to the sense

o agency, as well as the match

between orward model prediction and

outcome.

Why do we have a conscious

experience o agency? The

prospective aspect o agency gives a

clue as to whythe brain might provide

a distinctive conscious experience o

being in control. Selecting between

alternative action choices is oten a

dicult computational problem or

the brain. An experience o sense

o agency corresponding to fuent

choice may be an important signal

or guiding these computations.

For example, we oten make errors

by pressing the wrong button on a

machine, or by not knowing which

button to press. The prospective

sense o agency may be the brains

way o generating the eeling o just

knowing the right button to press.

In contrast, prospective eelings o

nonagency may allow time to slow

down, perhaps switching the brain to a

more attentive strategy o supervisory

control. It is clearly useul to have

such eelings prospectively,before

pressing the wrong button, yet existing

comparator models can only explain

the retrospective eeling o errorafter

one has pressed the wrong button.

The prospective sense o agencymay serve as a marker when action

processing needs to shit rom routine

to supervisory control.

We saw above that retrospective

sense o agency can be tricked, and

is sometimes illusory. The prospective

sense o agency may also be an

illusion. Simply having a eeling o

fuently knowing which action to

select does not guarantee the correct

action outcome. For example, I may

clearly eel that I know which button

to press on the machine, but I may

actually press the wrong button, orthe machine may malunction. The

prospective sense o agency might

only develop once the brain has

learned a stable relation between

actions and outcomes.

Where can I nd out more?Chambon, V., Wenke, D., Fleming, S.M., Prinz,

W., and Haggard, P. (2012). An online neuralsubstrate or sense o agency. Cerebr. Cortexin press.

Farrer, C., Frey, S.H., Van Horn, J.D., Tunik, E.,Turk, D., Inati, S., and Graton, S.T. (2008). Theangular gyrus computes action awarenessrepresentations. Cerebr. Cortex 18, 254261.

Johansson, P., Hall, L., Sikstrom, S., and Olsson, A.(2005). Failure to detect mismatches betweenintention and outcome in a simple decisiontask. Science310, 116119.

Spence, S. (2009). The Actors Brain: Exploring theCognitive Neuroscience o Free Will (Oxord:Oxord University Press).

Wenke, D., Fleming, S.M., and Haggard, P. (2010).Subliminal priming o actions infuences senseo control over eects o action. Cognition 115,2638.

1Institute o Cognitive Neuroscience,

University College London, London,

UK. 2Current aliation: Laboratoire de

Neurosciences Cognitives, INSERM,

Ecole Normale Suprieure, Paris.

Email:[email protected]

Qualia

Ryota Kanai1,2

and Naotsugu Tsuchiya2,3

Perhaps the most dicult biological

question o all might be how and why

electrochemical neuronal activity

in the brain generates subjective

conscious experience such as the

redness o red or the painulness o

pain. Neuroscientists track how light

impinging on the retina is transormed

into electrical pulses (neuronal spikes),

relayed through the visual thalamus

to reach the visual cortex, and nally

culminates in activity within speech

related areas causing us to say red.

But how such experience as theredness o red emerges rom the

processing o sensory inormation is

utterly mysterious. It is also unclear why

these experiences possess phenomenal

characteristics, which can be directly

accessed only rom the subject

having the experience. This is called

the hard problem o consciousness

as coined by the philosopher David

Chalmers. The phenomenal aspect

o consciousness or what it is like

character o subjective experience

is called qualia; the singular orm o

the word is quale, rom the Latin orwhat sort or what kind. In this Primer,

we provide an overview o the term

qualia and its conceptual issues, and

how neurobiological approaches can

contribute to clariy some o these

issues.

The diculty with qualia is their

subjective nature: qualia exist only

as viewed rom the inside. They

cannot be objectively detected or

compared like any other properties

measured in natural sciences. The

subjectivity o qualia allows one to

conceive hypothetical situations

that philosophers discuss in thought

experiments. For instance, we can

assume, without any contradiction, a

person withinverted qualia who, when

seeing the colour red, has the quale

that you would have when seeing

the colour green, and conversely.

In another thought experiment

we can conceive aphilosophical

zombie who has all the cognitive

and perceptual abilities like us but

lacks any phenomenal experience.

These thought experiments are oten

Primer
mailto:[email protected]:[email protected]:[email protected]

haggard & chambon 2012 sense of agency

Documents