Studying Cognition in Animals
Cognition
 Cognition: study of the internal states
and processes that produce behavior.
 Consciousness: something else
entirely.
Intentionality
 “I’ll pick up the kids from school today”,
says Max leaving for work. We’d normally
say that Max’s statement conveys an
intention. We can predict that he will drive
from his office to the school by a certain
route at a certain time and that he will
change his behavior if the circumstances
change.
 A philosopher might say that Max exhibits
intentionality.
Intentionality II
 Intentionality in the philosophical sense is
the property of being “about something”.
 It is the defining property of mental states.
 Beliefs, desires, plans, understandings, as
well as intentions, are examples of
intentional states.
 So, a belief has to be about something.
Hierarchy of Intentionality
 Zero-Order Intentionality: an animal that
does not have beliefs, desires, or any
intentional state, is exhibiting zero-order
intentionality.
 First Order: an animal with beliefs and
desires about the real world or the behavior
of others is a first-order intentional system.
 Second Order: When an animal’s mental
states concern the mental states of others.
 Thus, if Max plans to arrive at the school on
time, he is exhibiting first-order
intentionality.
 If he believes the children know he is
coming, he is exhibiting second-order
intentionality.
 If he wants them to believe that he expects
them to be waiting, then he is exhibiting
third-order intentionality.
Practical question:
 How do you measure this in an animal?

A hallmark of intention is that flexible behavior
is used to achieve a goal.
Broken-wing display in plovers
 E.g., Ristau’s work
 First order: plover plans to lead the fox
away from the nest

Second order: plover wants the fox to think she
is easy prey or has no nest
Evidence from plovers
 Flexible behavior: In 87% of staged encounters with a
human, plovers moved in a direction that was away from
the nest.
 Knowledge of other: plovers moved further away for
“dangerous” intruders than “nonthreatening” intruders
 Should monitor intruder: starts display when intruder can
see it, if the intruder stops following, plover intensifies
display, and approaches intruder.
 But can more hard-wired behaviors (ethological
approach) explain these changes in behavior?
 Sign stimuli and vectors? Eye direction?
 Series of if/then statements based on combinations
of sign stimuli?
Morgan’s Canon:Accept the lowest
level of intentional explanation that works
 From the point of view of natural selection,
what matters is that the animal achieves
goals such as finding food, mates, and
safety.
 Using Morgan’s canon to choose among
alternatives assumes that natural selection
has always produced the lowest level
intentional system that can do the job.
 Do you all agree with this?
Why is this relevant?
 Functional concerns of behavioral
ecology and ethology often lead to
mechanistic questions -- which are
the realm of cognition.
 Cognitive ethologists are frequently
concerned with the diversity of
solutions that living organisms have
found for common problems.

They also emphasize broad taxonomic
comparisons and do not focus on a few
select representatives of limited taxa (we
hope).
Darwin
 Cognitive ethology
can trace its
beginnings to the
writings of Charles
Darwinan

and some of his
contemporaries and
disciples.
The Expression of Emotion in
Animals and Man (Darwin, 1872)
 Argument against The
anatomy and physiology of
expression, by Bell,

Bell claimed that humans
were unique in their capacity
for expressing their feelings
through facial muscles.
 Darwin refutes the old
natural-theological
position,

through argument

and expensive foldout
heliotype illustrations!
Cognitive Ethology
(Griffin, 1976)
 The Question of Animal Awareness:
Evolutionary Continuity of Mental
Experience (Donald R. Griffin, 1976).
 Concentrates on the evolution and
evolutionary continuity of animal cognition
 Griffin wanted to come to terms with the
difficult question of "what is it like to be a
particular animal?"
Griffin’s approach





Mental images
Intentionality
Language
Communication
Behavior is plastic, modifiable, and
planning seems to be evident.
 Example: Bees communicate symbolically
with meaning (depends on map, who is
‘listening’, etc).
Summary of Critiques
 Problem: it is suggestive but not indicative



one cannot make unobservable mental experience
the heart of a field
few theories with testable predictions
ignores methods of cognitive science because they
are too “restrictive”.
 Speaks of mental experiences of animals
(“knowing that they think”).

Concludes it must be cognition when it’s actually
metacognition.
Solutions...
 Instead ask: how is information acquired,
processed, and used?


not “does it exist?” (mental experience).
make mental events, not experiences, the heart
of the field.
• You don’t have to experience the events for them to
be real and cognitive (the experience is
metacognition).
• Gives researchers something about which they can
make predictions and measure.
Information-processing method
Stimulus
States
and
Processes
Response
 Cognition: internal
states and processes
that produce behavior.
 This is tough: those
states and processes
are not directly
measurable.
 Do we need physicality of
structures to account for
empirical phenomena at a
behavioral level?
 No, not if one uses strong
inference.
Problem #2
 Natural selection acts
on outcomes (not
mechanisms).
 You can have > 1
mechanism and the
same outcome or

Spatial Memory in Corvids
…>1 outcome with the
same mechanism.
 You need to constrain
the question. How?
 Ecology!
Ecological Approach to
Cognition
Environment
Transformed
by...
Cognitive
Structures
 Assume: function
constrains mechanism.
 Correspondence between
Behavior
mechanism and function
to
may not be neat.
predict

How is info here…
Studying ecology should
give us clues.
Behavioral ecology can inform
questions of cognition
 Optimal Foraging Theory: maximize rate of energy intake
and fitness.
 Example: Woodpecker takes longer on some trees than on
others when foraging.
 Assume: adaptive, optimal, maximizing energy, maximizing
fitness.
 Function: avoid depleted food, avoid predation, stay close to
nest.
 Mechanism: how know depleted, what info tells bird to change
behavior, how does it know where its nest is?
 Measure: distances between trees, prey repletion rates, prey
energy, etc.
Ex) Kinds of Memory
 Reference:
Information that is
procedural and long
term.

How to dial phone
 Working: event
specific and short
term.

What number to dial.
Animal patrolling a territory
 Reference Memory:
Domain of territory and
how to patrol
 Working Memory: relevant
events of last patrol
(intruders and ripening of
food).
 Amakihi anecdote and
study…
 …avoids mist net and
drives intruder into it.

spatial memory?
Learning and Foraging
 Much of what goes on in foraging
(MVT, OFT, etc) implies learning
and memory
 Operant conditioning even!
 Amakihi: Always visits renewed
nectaries in flowers on territory
(maximizes energy intake).
 Mechanism: Must remember
 1) where flowers are

2) the last time visited
 Measure: levels of nectar, time
between visits, energy intake rate,
spatial efficiency.
 But do this randomly so an intruder can’t figure it
out!
??
Food Storing Behavior
 You create a resource distribution only
you know.
 Reference Memory: storage sites, what is
in the site, territory
 Working Memory: which site did I empty
today?
 Information: spatial layout, site contents,
etc.
Spatial Memory: Corvids
Ecology
1) Where is food?
2) What food is it?
Cog. Struct
Mechanism
Offers a unique opportunity
to use the comparative
approach…another dimension
to add.
Behavior
1) Storage place
2) Food type
The Synthetic Approach to the
Study of Animal Cognition
 3 Major Aspects



1) Broad definition of the phenomena of
interest
2) Assume and use comparative evolutionary
origins
3) Emphasis on the importance of studying
learning in the laboratory and in the field
• Multiprocedural (to deal with L/P distinction)
• Internal and External Validity
Broad Definition
 Intelligence: Processes by which animals
obtain and retain information about their
environment and use that information to
make behavioral decisions.



Cognitive AND adaptive
Integrates nature and nurture
NOT general
Comparative Evolutionary
Approach
 Anagenesis: linear ranking of species or a
trait.


No, no, no….
Please read Hodos and Campbell every year
(1969)
 Note: both quantitative and qualitative
differences are of interest.
Laboratory and Field
 If intelligence serves an adaptive function,
one must look at environmental constraints.
Cognitive
Structures
Transformed
by...
Environmental Info
Behavior
to
predict
Laboratory/Field Studies
 important for two reasons
 1) External Validity: if questions from the
laboratory work in the field= RIGOR
 2) Theoretical Reasons: you’re looking at
how behavior affects biological success,
you must look at this in the field.
Biological Approach II
 Assumes there MUST be significant
variation in animal intelligence
 Feather analogy: feathers are so successful
birds have never lost them. But they are not
so successful that there is no variation at all.


Same goes for learning!
Traits with adaptive function will therefore
vary between species.
Field Studies
 Learning is more difficult to observe than
learned behaviors.




Seed caching birds (contents of caches)
Bees
Vervets
Foraging models (animals learn energy and
handling time and rank items).
Research Strategies

Develop natural history of animal intelligence
• foraging
• social behavior
• learning

Use natural history to choose species and
design procedures
• Field: Is there a gradient in a behavior in a set of
species?
• Ecological validity in lab: how does the task meet a
requirement of an equivalent problem faced in
nature
– ex) Blue Jays and cryptic prey
Research Strategies II
 learning/performance distinction versus true
species differences


Multiprocedural Approach: you cannot
eliminate all variables, so change the context
but keep the question the same (remember Glen
Hass’s experiments from last week)
Have external criteria to make predictions about
species differences
• This is better than a null hypothesis of NO
differences
• If all predictions are supported in all tests, this
argues against contextual variables.
• Base it on natural history
Generate multiple predictions
about species differences (Strong
Inference, Platt)
 Divergences: differences in a cognitive trait
between closely related species due to
ecology.
 Correlations: similarities in a cognitive trait
due to similar ecology.
 E.g., Spatial memory and amakihis.
Compare them to a close relative with
different feeding habits or to a nectar
feeding bat….better yet….
Really do a comparative study
 You need at least 3 comparisons (2 closely
related, and an out-group of some sort)
 Control for phylogeny AND ecology
 No cat, rat, chicken, and man comparisons!
So:
 Set up External Criteria
 Generate Multiple Predictions
 Use Multiprocedural Approach





Several procedures that measure the same
cognitive process
Test the 3 species with all the procedures
Choose species based on external criterion from
their natural history
Make predictions in advance
If the same order of species results from each
different experiment, the contextual variables
are unlikely to be responsible
So, part II:
 1) Broadly define learning, intelligence, and
cognition
 2) Place learning, intelligence, and cognition in an
evolutionary context
 3) Use behavioral ecology as a basis to study
learning, etc.
 4) Use the comparative approach:



Your hypotheses should allow you to make multiple
and detailed predictions
Natural history should help you generate your multiple
hypotheses
Multiprocedural approach
Example: Food storing in corvids




Gradient in Natural History
Gradient in Morphology
Gradient in reliance of food stores
Gradient in food-storing behavior
Mexican
Scrub
Pinyon
Nutcracker
Based on Ecology: spatial
 Predict: species differences in spatial
memory capabilities:
mexican<scrub<pinyon<nutcracker
 Multiprocedural Approach





Radial Maze
Food store and recovery (long term, short term)
Operant tests of short-term spatial memory
Differences in hippocampal volume
Species differences hold up in all tests, no
matter the context.
Sociality
 It has been suggested that the cognitive
demands of social living may be an important
force driving the evolution of intelligence.
 Kamil and colleagues have been testing this
hypothesis by comparing highly-social pinyon
jays (Gymnorhinus cyanocephalus) with their
less social close relative, the western scrub jay
(Aphelocoma callifornicus).
 Tests such as observational learning
(Templeton, Kamil & Balda, 1999) and
transitive inference (Bond, Kamil & Balda
2002) have been used.
Based on ecology:sociality
 Reference, nonspatial task


Predictions: scrub<<pinyon
Results support predictions.
 Now test against nonsocial Nutcracker
Scrub
Pinyon
Spatial Cognition
 On the small scale, Kamil and Jones were
interested in how Clark's nutcrackers
(Nucifraga columbiana) relocate their seed
caches.
 On a larger scale, they were interested in
nutcrackers' ability to learn geometric
relationships (Kamil & Jones, 1997, 2000) and
how nutcrackers recognize locations where
food has been stored (Kamil & Cheng, 2001).
Can the animal detect the information required to
construct a representation (e.g., geometric relations
between objects)?
 Clark’s Nutcrackers: Birds use general
principle (relationship between landmarks,
not between a goal and the landmarks) to
find a goal located between two landmarks.
2)
1)
Goal
Two Landmarks
 Clark's nutcrackers can learn to find the point halfway
between two landmarks that vary in the distance that
separates them.

A general principle, as the birds correctly find the halfway
point when the landmarks are presented with new distances
between them.
 The ability to find a point defined not by the
relationship between a goal and a landmark

but by the relationship between landmarks.
 Two distinct processes:

direction: the use of directional bearings to find the
(hypothetical) line connecting the landmarks and

distance: finding the correct place along that line.
 Nutcrackers were trained to find a location defined
by its geometric relationship to a pair of landmarks.

Two groups were trained to find positions on the
line connecting the landmarks

two groups were trained to find the third point of a triangle,
-- based on either constant directional or distance
relationships to the landmarks.
 Four inter-landmark distances and a constant spatial
orientation were used throughout training.
 Result: distance group learned more slowly with less
accuracy and showed less transfer to new distances
-- than did birds in the other groups.
 Finally, when tested with a single landmark

birds in the half and quarter groups tended to dig in the
appropriate direction from the landmark, as did birds in
the distance group.
 Nutcrackers can learn a variety of geometric
principles:

directional information may be weighted more heavily
than distance information

the birds can use both absolute and relative, including
configural, information about spatial relationships.
Three Landmarks
Hunting by search image
 Five forms (or
"morphs") of the North
American underwing
moth, Catocala relicta.
(Revised from Barnes &
McDunnough 1918).
 Note the variable forewings and the relatively
uniform hind wings.
 hunting by searching
image.
Stimuli
 Artificial moths on
artificial backgrounds.
Testing
 Operant
 Moth and no moth
trials
 Either peck moth or
key saying “no
moth”
Results
 Runs of the same type of
prey resulted in “search
image” effects
 Also interference effects:
-- making a jay to
search for one type of
moth actually reduced the
likelihood of its finding an
alternative type.
 This was the first clear
demonstration of
attentional interference in
visual search in animals.