tinbergen’s four questions for investigating...
TRANSCRIPT
• Tinbergen’s four ‚questions’ for investigating behavior• Mechanism
• Ontogeny
• Function
• Evolution
Topic for today
Socio-cognitive abilities of dogs – mainstream research direction is bottom-up
It starts with a comparative question: „Do dogs have the capacity of…, such as the humans/apes…?” e. g.
Following visual cuesLearning from demonstratorsAttribute false beliefs to others, etc.
Studies on vocal behavior however are usually based on observationsof commonly occuring phenomena, similarly to some other examples,like
Attachment behavior, empathy, ‚guilt’ etc.
Vocal behavior of dogs provides a unique chance for ethologists tostudy natural dog behavior and its possible functions, as well as an exciting opportunity for building theories about the course of evolution during the domestication process
Describing the abundance of canine vocalizationsEarly, but throughout efforts (Cohen and Fox, 1976)
Recurrent problems with naming the categoriesBetween-language correspondanceA new (acoustics-based) categorization would be necessary
Are there dog-specificvocalizations?
*
*= woof woof in Hungarian
Compared to the closestrelative(wolves – Feddersen-Petersen, 2000)
Different acoustics? (yes and no)Different contexts? (yes and no)
WHY?Function?Evolution?
Answer may shed light to
events during domestication
Cohen & Fox,
1976
meow
gru
nt
wh
ine
yelp
scre
am yip
ho
wl
co
o
gro
wl
co
ugh
bark
clic
k
too
th
snap
pan
t
Greeting F WD WC
D
D F C WD R WD - D - - F
Play
initiation
- - D D - - - - - - D - WK FD
Submission F - WC
D
D W
CD
- - - - - - - - -
Defense - - WC
D
K CF
W
- W - WC
DF
WC
DF
WD F WC
D
-
Threat - - - - - - - - WC
DF
WC
DF
WC
DF
F WC
D
-
Contact
seeking
nF DF n, D
W C
nW
nC D
F C - F - - D - - -
Pain/ distress n - n, D
WC
F
nW
nC D
n, D
W C
F
- - - D - D - - -
Loneliness - - n, D
WC
D - - W C
D
F - - D - - -
Group
vocalization
- - n, D
W C
- - C W C
D
- W
C D
- - - - -
Predominantly dog-specific occurrence
Abundant (easy to experience/collect)
Large acoustic variability, and different from the wolves’ bark
The role of barks in dog-dog communication
Early ‚non-communicative’ theoriesHypertrophyExcitementNeoteny
Mobbing theory (Lord et al., 2009)
Acoustic similarities (bark = mobbingsignal)
Bark contexts are only partly overlappingwith the opportunity to mob (unlikebarking alone, joint action with owneretc.)
Cohen and Fox, 1976
Coppinger and Feinstein, 1991
Back to the basics – reliable acousticfeatures are needed for possiblecategories of informationIdentity, affective state, context, etc.
Early analyses (few parameters, related tomostly context and affective state)Yin, 2002; Pongrácz et al., 2005)
Machine learning approaches (manyparameters, also about identity, sex, age of the dog)(Molnár et al., 2008; Larrañaga et al., 2013 – Anim. Cogn.)
The acoustic basis is given
for information transfer
between dogs
Habituation-dishabituation paradigmDogs show evidence of ‚hearing the difference’
BUT the circumstances were far from being realisticWhat about the natural behavior?
IDENTITY, CONTEXT
Maros et al., 2008 –
Appl. Anim. Behav. Sci.
Molnár et al., 2009 –
Behav. Proc.
The effects of context (affectivestate) and identity were tested
Territorial (‚stranger’) barks elicitstrong reaction
Distress (‚alone’) barks do notFamiliarity of the caller also counts
PARTICULAR types of barks
elicit appropriate response
from the receiver dog
OTHER barks (‚alone’) elicit
no response
These latter are ‚new’
compared to the wolf barks
If they are communicative
signals, who is the
AUDIENCE?
Pongrácz et al., 2014 – Appl. Anim. Behav. Sci.
ChemicalWith little to no relevance for humans…
VisualDogs understand a variety of human visual cues
(Lakatos et al., 2009)Humans have to learn the body language of dogs
(sometimes difficult)Misunderstanding can be fatal (small children / aggressive face expressions of dogs – Meints et al.. 2010)
Humans understand ‘humanized’ visual cues, like gaze alternation (Miklósi et al., 2000)
AcousticDogs are usually very vocal, but do we understand
them? ??
Selection pressure alone is not an explanation itobviously did not work in the case of chemical and most of the visual signals
‚Universal’ channel for (even interspecific) communication?
Two main explanatory hypothesesEvolution structural-motivational (Morton, 1977)Mechanism source-filter (Taylor and Reby, 2010)
General approach – playback studieswith large amount of standardizedsound recordings (from Mudi dogs)
CONTEXTSStranger at the gateSchutzhund trainingLeft aloneBefore walk‚Asking’ for ballPlay
Tasks for the participantsContextual recognitionRating of EMOTIONSAggressiveness, Fear, Despair, Happiness, Playfulness
0
5
10
15
20
25
30
35
40
45
50
Mudi-owners Dog-owners Non-owners
rec
og
niz
ed
sit
uati
on
s (
% +
SE
)
chance level=16.7%
***: P<0.001
*** *** ***
Adults with different amount of experience with dogs
High overall success rate of context recognition
No effect of experience (Pongrácz et al., 2005 – J. Comp.
Psychol.)
Sighted and sightless adults (visual experience varies)
High overall success rate of context recognition
No effect of sightedness (Molnár et al., 2010 – Q. J. Exp. Psychol.)
0
5
10
15
20
25
30
35
40
45
Sightless from birth Sightless with visual
experiment
Sighted
co
rre
ctl
y c
ate
go
rize
d s
itu
ati
on
s (
%)
chance level: 17%
*** *** ***
Children with different extent of dog-experience
Only three contexts, success rate mostly above chance level
No effect of experience, general effect of age (Pongrácz et al.,
2011 – Appl. Anim. Behav. Sci.)
0
10
20
30
40
50
60
70
80
90
100
6 years
old
8 years
old
10
years
old
adults
corr
ectly
cat
ego
rize
d s
ituat
ion
(%+S
E)
dog yes
dog no* NS * *** *** *** *** ***
Experience had only minimal effect
Overall recognition rate is well above chance level
Strong effect of context on recognition accuracy
0
10
20
30
40
50
60
70
Stranger Schutzh. Walk Alone Ball Play
reco
gn
ized
sit
uati
on
s (
% +
SE
)
a ac bc ac bc c
chance
level=16.7%
Pongrácz et al., 2005 – J. Comp. Psychol.
What could be the reason?
Pongrácz et al., 2006
Silva et al., submitted
Repeated study on
Hungarian and
Portuguese samples.
Affective state of dogs in particular contexts was rated ratheruniformly across the sample‚stranger’, ‚Schutzhund’‚alone’‚play’
These contexts were easier to recognize contextually as well
Emotional assessment Contextual recognition
(unconscious) (conscious)
Wolf-like ancestor
• Barks encode• Aggression
(defensive/offensive)
• Other affective statesare being encoded by• Howling
• Growls
• Whining, etc.
Dogs
• Barks encode• Aggression
(defensive/offensive)• Fear• Frustration• General excitement (?)• Happiness / positive
affective state
• Other types of vocalization remain intact, or become more rare(e.g. howling)
What kind of selective pressure?
Human environment is of crucial importance
Protection (i.e. passive tolerance) against thenegative consequences of being loud (both thepredators and shortage of prey)
Preference for particular functions of barking(alarm, hunting, herding)
Some forms of barks were retained (aggression)Some new forms turned to be useful (fear/separation)Some new forms were tolerated (playfulness)
But see the new challenge!
‚Nuisance barks’
‘Theory of communicative relevance of auditory nuisance’
Particular signal-structures evolved for intenseattention elicitingBaby cries (artificial version: siren)
Natural response: interventionDog barks do the sameWhen intervention is impossible…
STRESS ANNOYANCE
High pitched barks are the most annoyingAttention eliciting (baby cry, cat meow, sirens)No habituationTypical to dogs in trouble, separation
Fast, harsh, deep barks are very annoyingAggressive vocalizations‘Original’ function of barks – similarity to wolvesFight, flee, or intervene
‚Works’, because of the NEW FUNCTION of barks
‚Works’, because of the OLD FUNCTION of barks
Growls – apparently no ‚new function’ in the human
environment
Agonistic interactions
Play
Commonly encountered in dog-dog and dog-human
interactions
Testing of the information content of growls
Step 1: intraspecific relations
Step 2: dog-human communication
Playful vs. two types of agonistic growls
Test situation matches with only one of the
contexts
Dogs gave up only when a
‚food guarding’ growl was
played back
Discrimination
between agonistic and
playful growls?
Discrimination
between two types of
agonistic growls?
Recognition of the
food guarding growls?
functional referentiality?
Faragó et al., 2010, Animal Behaviour
2
7
31
3
4
9
2
5
0123456789
101112
FG TS PL
Type of growl playback
leave
eat
take
‚Human version’ of the
aforementioned study
Choose the correct context
Rate the affective state
Each context was recognized
above chance level
Emotional rating congruent with
context
Food guarding growls were the
most ‚aggressive’ for HUMANS
Maybe for dogs, too?
(Faragó et al., subm.)
Formant dispersion of play and
food guarding growls from the
same dog show interesting
difference
Formant dispersion – connected
to the size of the vocal apparatus
indirectly to body size
Do dogs ‚manipulate’ the
indexical information about
themselves?
A A
B950
1000
1050
1100
1150
FG TS PL
Formant dispersion
Play growl
Food guarding growl
Faragó et al., 2010, PLoS ONE Bálint et al., 2013, Appl . Anim. Behav. Sci.
Food-guarding growls
Dogs look at the
‚matching’
size DOG picture
Playful growls
Dogs look at the
‚bigger’
size DOG picture
+ sound playback
The honesty of agonistic growls may originate from theclose-distance circumstances of these vocalizations, where cheating would be difficult due to the instant visual reference
Exaggerated size information encoded in play-growlsmay serve to sustain playful context due to theobvious discrepancy between the acoustic and visualinformation
What about indexical signaling towards humans?
Taylor et al. (2008) found in play-back studies that
humans are sensitive to the size-component of dog growls
Would dogs use ‚honest’ signals in an agonistic encounter
also towards HUMANS?
‚Threatening stranger’
experiment
Various human
participants of ‚small’
and ‚large’ body size
Male
Female
Multi-parameter
acoustic analysis of the
elicited growls
Dog growls with lower
fundamental frequency and lower
formant dispersion at
MEN
Experience counts, the effect is
the strongest if the dogs lived in
MIXED-GENDER families
First time evidence was found
that dogs modify vocal indexical
information as a function of
different
AGONISTIC encounters
Bálint, Faragó, Miklósi, Pongrácz (2016) in press Anim. Cogn.
moan
whine growl
whimper
snarl woofyelp
Fear, pain
Play
Warning, defense
Frustration, anxiousness
Tonal sounds Atonal sounds
bark
Repetitiveness grows
Tonal barks
High pitched barks
Recruiting effect;
Long distance communication
Non-aggressive
inner states encoded
Long distance
non-aggressive communication
Hypothesized receivers:
HUMANS
Growls convey an abundance of information to other DOGS
Growls are also understandable for HUMANS
inner state ? context(indexical features) body size
Growls were found suitable for
dynamic apparent size communication
Voluntary control???Learning???
Effect on the receiver???
Evolution
Did breed-selection affect vocal behavior?
Ontogeny
How does the social environment affect the
emergence of particular vocalization types?
Mechanism
Are there specific acoustic ‚fingerprints’ of
conditions such as separation-related disorder?
(see the lecture of Faragó onWednesday morning)
Function
What type of information do particular types of
vocalizations convey to humans and dogs?