Animal Cognition

Question 1

1. Cognition and the Study of Behavior

Studying cognition means analyzing how animals acquire, process, and use information.

Answer 1

Walnut trees shade the streets of Davis, California. They also provide food for the crows that roost near Davis. Crows crack walnuts by dropping them from heights of
5–10 meters or more onto sidewalks, roads, and parking lots. Occasionally they drop walnuts in front of approaching cars, as if using the cars to crush the nuts for them. Do crows intentionally use cars as nutcrackers?

A team of young biologists at UC Davis put this anecdote to the test (Cristol et al. 1997). They reasoned that if crows were using cars as tools, the birds would be more likely to drop nuts onto the road when cars were coming than when the road was empty. Furthermore, if a crow was standing in the road with an uncracked walnut as a car approached, it should leave the nut in the road to be crushed rather than carry it away. They found no support for the notion that crows were using automobiles as nutcrackers. In other respects, however, the birds’ behavior with walnuts was quite sophisticated (Cristol and Switzer 1999).

For example, by dropping nuts from buildings on the Davis campus, Cristol and Switzer verified that English walnuts did not have to be carried so high before breaking as the harder black walnuts and that they broke more easily when dropped onto pavement than onto soil. The crows’ behavior reflected these facts. A crow dropping a nut also took into account the likelihood that a greedy fellow crow might steal a dropped nut before it could be retrieved: the fewer crows waiting on the ground nearby, the higher they took walnuts before dropping them.

The story of the nutcracking crows encapsulates some key issues in the study of cognition in animals. Foremost is how to translate a hypothesis about essentially unobservable internal processes into hypotheses about behavior in a way that permits
different explanations to be distinguished. This meant asking, ‘‘What will crows do if they are using cars as tools that they will not do if they are merely dropping nuts onto the road as a car happens by?’’

A second issue has to do with the kinds of hypotheses people entertain about the processes underlying animal behavior. The people in Davis and elsewhere (Nihei 1995; Caffrey 2001) who saw nutcracking as an
expression of clever crows’ ability to reason and plan were engaging in an anthropomorphism that is common even among professional students of animal
behavior (Kennedy 1992; Wynne 2007a, 2007b).

research often reveals that simple processes apparently quite unlike explicit reasoning are doing surprisingly complex jobs. Free-living crows were observed doing something suggestive of interesting information processing and decision making. Their behavior was then examined with more systematic observations and experiments. Among other things, these revealed how closely the crows’ behavior matched environmental requirements.

Numerous cognitive processes underlie the crows’ nutcracking, and each of these could be analyzed further. For example, how do crows judge the height from which they drop nuts? Do they have to learn to adjust their behavior to the kind of nut, the kind of substrate, and the number of nearby crows? Several species of other birds break hard-shelled prey by dropping them (Cristol and Switzer 1999), and one might also ask what ecological conditions or evolutionary history favor this behavior.

Question 2

1. Cognition and the Study of Behavior
2. 1. What is comparative cognition about?

Most people who study comparative cognition remain agnostic as to whether animals process information consciously or not. Some animals may be conscious in some sense, but we cannot know because consciousness refers to a private subjective state. Furthermore, it is often difficult to specify any behavior uniquely resulting from consciousness. How animals process information and behave adaptively can be understood, and on the whole should be studied, without reference to consciousness. Nevertheless, some researchers are of the opinion that some animals are undoubtedly conscious, and scientists should be trying to understand the nature of their conscious
states.

Answer 2

1.1.1 What is cognition?
Cognition refers to the mechanisms by which animals acquire, process, store, and act on information from the environment. These include perception, learning, memory, and decision-making. The study of comparative cognition is therefore concerned with how
animals process information, starting with how information is acquired by the senses. Cognitive is often reserved for the manipulation of declarative (knowing what) rather than procedural (knowing who or knowing what to do) knowledge (e.g., Dickinson 2008).

The first kind of representation implies more flexible behavior than the second, but in both cases behavior results form processing and storing information about the world. First-order processes operate directly on perceptual input, as when a stimulus triggers a response or creates a trace in memory. Second-order processes (regarded as interestingly cognitive) operate on first-order processes, as in evaluating the strength of one’s memory for an event (Heyes 2008; Penn, Holyoak, and Povinelli 2008).

It is almost never possible to tell without experimental analysis what kinds of processes are reflected in a given behavior (Dyer 1994). Comparing the ways in which different species solve similar information-processing problems is an important part of the comparative study of cognition, it should embrace all sorts of information processing and decision-making.

1.1.2 Animal cognition or comparative cognition?

Some classic assessments of psychological research on animals (Beach 1950; Hodos and Campbell 1969; Dewsbury 1998) are complaints that most studies labeled ‘‘comparative’’ are mere ‘‘animal psychology’’
because they deal with only a single nonhuman species or at most implicitly compare that one species with humans.
Thorough analyses of cognitive processes in limited species form the foundation for comparative work, as when comparisons of memory in food storing and nonstoring birds draw on method and theory developed in studies with pigeons.
Therefore ‘‘animal cognition’’ research is part of the overarching enterprise referred to as research on comparative cognition aimed at understanding cognition across the animal kingdom, including how it works, what it is good for in nature, and how it evolved.

why study animal cognition?

• understanding more about ourselves through comparisons (simularities/differences, what other’s don’t do)
• understanding about that organism
• making inferences about evolution

1.1.3 Consciousness and animal cognition

People intuitively distinguish between merely responding to events and being aware of them. Perceptual awareness can be distinguished from reflective consciousness (reflecting on our thoughts and actions). One impetus for this work was the discovery of ‘‘blindsight’’ and priming in memory, which reveal distinct conscious and unconscious processes in everyday cognition.

Debates about the extent to which people are aware of their own cognition have also placed a new emphasis on how subjects consciously experience their memories, percepts, or the like as distinct from how they act on them. Progress in analyzing the neural basis of behavior in such experiments through brain imaging and studies of cognitively impaired people
have encouraged attempts to investigate the same processes in animals (e.g., Terrace & Metcalfe 2005).

A central methodological problem here; evidence for consciousness in humans generally consists of what people say about their mental experiences, seeking it in nonverbal species requires us to accept some piece of the animal’s behavior as equivalent to a person’s verbal report. Therefore, the point of view of most researchers studying animal cognition is that how animals process information can, and should, be analyzed without making any assumptions about what their private experiences are like (Staddon 2000; Hampton 2005; Heyes 2008).

This approach takes support from evidence that people act without being aware of the reasons for their actions (without using reflective consciousness). A related view
(Macphail 1998) is that human babies nor nonhuman animals can have reflective consciousness because it requires language. The view that consciousness in animals is not a subject for research either because
it is inaccessible to scientific study or because animals lack language was emphatically rejected by scientists calling themselves cognitive ethologists (Ristau 1991a).

Stimulated by Donald Griffin (1976, 2001), cognitive ethologists claim that much behavior suggests that animals have intentions, beliefs, and self-awareness, and that they consciously think about alternative courses of action and make plans (Griffin and Speck 2004). Studies of animals communicating, using tools, and apparently deceiving one another seem to reveal flexible behaviors governed by intentions to achieve specific goals. However, it is difficult to find a
situation for which the notion that an animal has a conscious belief or intention or is consciously manipulating information unambiguously predicts what it does (Dawkins 1993; but see Griffin 2001).

Nevertheless, the early years of the 21st century some anthropomorphic mentalistic terms have traditionally been accepted to refer to processes underlying animal behavior. For example, training a rat that a tone predicts a shock is usually referred to as fear conditioning which indeed may be in the same physiological state as a person describing himself as fearful. Similarly, a hungry rat trained to press a lever for food could be said to be doing so because it desires food and believes that lever pressing will give it food.

On one view (Dickinson 2008) the goal-directedness of
bar pressing or other instrumental responding, that is, evidence that it is controlled by belief and desire, is what is meant by it being under cognitive control. Belief, desire, fear, or other mental or emotional states may be ascribed to animals on the basis of well-defined behavioral criteria, that is, on the basis of functional similarity, without implying that the animals are undergoing humanlike conscious experiences.

If we accept that human beings are conscious it seems that some other species, perhaps among primates, must share at least perceptual awareness with humans (see Terrace and Metcalfe 2005). Saying that only humans are conscious in any way seems like rejecting evolutionary continuity (but see Penn, Holyoak, and
Povinelli 2008). However, because evolution has acted via the results of what creatures do, not directly on what they experience privately while doing it, it seems there must be something promoting survival and reproduction that a conscious animal can do and one lacking consciousness cannot, but so far there are no clear candidates for that ‘‘something’’ (Dawkins 1993, 2006).

This same problem of an apparent evolutionary gap between humans and other living species arises in
discussions of the evolution of human language and abstract conceptual abilities (Penn, Holyoak, and Povinelli 2008). Despite the apparent successes
of teaching aspects of language to apes, most would now conclude that language is unique to humans, and the conditions under which it could evolve are an active area of debate. Anthropological studies of human evolution and of primate behavior in the wild are likely to add fuel to these discussions for some time to come.

1.1.4 A word about intelligence

It is sometimes said by cognitive ethologists (Griffin 1992) and popular writers (e.g., Barber 1994) that animals must be thinking because they behave so intelligently. Indeed, to the nonspecialist one of the most persuasive arguments that animals think
as we do is that it is impossible to imagine another explanation for their ‘‘clever’’ behavior (Blumberg and Wasserman 1995; Wynne 2004a). On the whole, however, intelligence is not a useful term for describing animal behavior, for two reasons.

First, intelligence is generally used to describe global ability in people, whereas the cognitive abilities of animals (and perhaps people as well) are to a large extent modular. For instance, a Clark’s nutcracker that can retrieve thousands of pine seeds months after caching them is not necessarily ‘‘smart’’ in other ways. It is particularly good at encoding and retaining certain kinds of spatial information, but it may remember other kinds of information no better than other birds.

Within this context, intelligence is sometimes used nowadays to refer to the collection of specific
cognitive abilities that a species may have (cf., Emery 2006; Pearce 2008). A second reason to use intelligence carefully is that it should be defined formally with respect to a specified goal (McFarland and Bosser 1993). On this view, biological intelligence should be defined in terms of fitness (Kacelnik 2006) or goals such as choosing a good mate that contribute to fitness, and even plants can be intelligent (Trewavas 2002). Sometimes intelligent behavior may be produced by very ‘‘unintelligent’’ means.

Finally, recent demonstrations that species differ in behavioral flexibility, or propensity to adopt novel foraging behaviors, have revived discussions of overall animal intelligence (cf., Roth & Dicke 2005). This is especially so because the correlation of flexibility with overall brain size and/or size of the forebrain in some animal groups satisfies the everyday equation of intelligence with ‘‘braininess.’’

The naively anthropocentric nature of such discussions is underlined by a comparison of pigeons and people in a test of complex reaction time (Vickrey and Neuringer 2000). In such a test the subject is confronted with an array of lights; a randomly chosen one, the target, lights up on each trial and the subject’s task is to touch it as quickly as possible. People with high IQ show the smallest increase, this is because high IQ reflects a general ability to process information fast. Therefore, ‘‘less intelligent’’ species should be affected more strongly by increasing numbers of targets than humans.

however, pigeons show a smaller effect than very intelligent humans tested in the same way. As the authors of this study observe, ‘‘the counterintuitive
conclusion follows that pigeons are more intelligent than people. An alternative view assumes that different intelligences or factors are employed in different
situations by

Question 3

1. Cognition and the Study of Behavior
2. 2. Kinds of explanation for behavior (how to explain cognition)

Four questions, often referred to as Tinbergen’s four whys, can be asked about any behavior. These are questions about immediate causation, development in the individual, present-day function, and evolution. The four questions are complementary; each contributes to a complete understanding of behavior. Cognitive mechanisms such as perception and memory are among the immediate causes of behavior; learning is part of behavioral development. Cognitive processes are also part of an animal’s adaptation to its environment and therefore must have evolved through natural selection.

Answer 3

1.2.1 Tinbergen’s four questions

The pioneering ethologist Niko Tinbergen (1963) emphasized that the question, ‘‘Why does the animal do that?’’ can mean four different things, sometimes referred to as ‘‘Tinbergen’s four whys.’’

• How does the behaviour occur in an individual? (causation) - proximate mechanism
• how does the behaviour arise in an individual? (development) - proximate ontogeny
• why is this behaviour adaptive for the species, what is the behavior good for; what is its survival value? (function) - ultimate adaptive value
• how does the behaviour arise in the species (evolution) - ultimate phylogeny

However, it is important to be clear on how they differ from one another and to avoid confusing the answer to one with the answer to another.

• The proximate cause of nut dropping would be sought in some interaction of the bird’s internal state, most
  likely hunger. as well as the presence of nuts, and of hard surfaces.
• young crows do not crack nuts but older crows do, which suggests that they learn through experience.
• Cracking walnuts clearly functions in obtaining food, but questions about the function of the crow’s behavior can also be asked at finer levels of detail. for example, why they carry the nut so high and not higher.
• ‘‘How did it evolve?’’ usually has to be tackled by
  trying to look at the behavior’s phylogenetic history using methods. comparing different species and groups

observing evolution happening in the wild ;
Rosemary Grant and Peter Grant observed the beaks of seed-eating finches on the Galapagos Islands
changing in response to drastic changes in rainfall (Grant and Grant 2008). In years of drought, only the birds most skilled at cracking the few remaining seeds could survive and reproduce. Beak depth, an indication of seed-cracking power, contributed importantly to survival in the medium ground finch (Geospiza fortis). Because beak depth is heritable, changes in the population’s distribution of beak depths could
be detected in a few generations. The birds’ behavior must have changed, too, perhaps through learning. Rather than ignoring the hardest seeds, as they did in
times of plenty, the successful individuals evidently became skilled at finding and cracking them.

1.2.2 ‘‘Learned’’ and ‘‘innate’’ behavior

No behavior is either strictly learned or entirely innate. An excellent illustration of how preexisting selective processes in the animal interact with specific experiences to produce learning comes from classic comparisons of song learning in two species of sparrows (Marler and Peters 1989). Like many other songbirds, male song sparrows (Melospiza melodia) and swamp sparrows (Melospiza georgiana) need to hear species-specific song early in life in order to sing it when they mature.

The two species are closely related, but swamp
sparrows sing a much simpler song. Marler and Peters played song sparrow songs and swamp sparrow songs to isolated young males of both species in the laboratory. Thus, all the birds had the same acoustic experience. But their behavior as adults revealed that they had learned different, species-appropriate things from it. Both sparrows had a strong preference to their songs.

In the wild, these two species may live within earshot of each other, so early-developing selectivity in perception and/or learning likely functions to ensure
that each one learns only its own species song (Marler and Peters 1989). a very important general point: cognitive mechanisms are adaptations to process and use certain kinds of information in certain ways, not
mechanisms for information processing in general. it implies that genes can work without an environment to work in, the term innate is never appropriate in modern biology (Bateson and Mameli 2007).

Hogan (1994b) has suggested the term prefunctional because it does not imply that the genes have worked in isolation nor that prior experience is irrelevant. However, because this term implies that we know the function of the behavior, predisposition or preexisting bias may be preferable. Finally, to say that some behavior or cognitive process develops prefunctionally is not to say that it is unmodifiable (Dawkins 1995). As the comparison of song and swamp sparrows illustrates, how much and in what ways behavior can be modified itself reflects events earlier in development.

This example also shows how a stereotyped behavior seen in most normally developing members of a species can result from learning. However, although it makes a key point in showing how experience can have species-species effects, it misleadingly implies that effects of experience (here, the songs) and genes (the species of sparrow) can always be neatly separated. Developmental biologists are increasingly documenting gene by environment interactions and interdependencies as well as epigenetic effects, in which environmental effects on the genes of one generation are passed on to the next (Sokolowski
& Levine in press). Some of these discoveries have implications for behavior; undoubtedly more such will be uncovered in the future.

In conclusion, structure as well as behavior, the animal’s phenotype, results from a continuous and seamless interplay of genes and environment that is itself selected. The extent to which behavior patterns or cognitive capacities are modifiable by experience varies so much as to make the terms learned and innate (or nature and nurture) obsolete (Bateson and Mameli 2007). The fact that individuals within a species (i.e., with a common genotype) may develop different physical and/or behavioral phenotypes in different environments is known as phenotypic plasticity. The
ability of individuals to learn details of their own environment that are unpredictable on an evolutionary timescale is but one aspect of the more general phenomenon of adaptive phenotypic plasticity (Dukas 1998; for a brief review see Agrawal 2001; for
further discussion of the topics in this section see Marler 2004).

Question 4

1. Cognition and the Study of Behavior
2. 3. Approaches to comparative cognition

Cognition in nonhuman species has traditionally been approached differently by psychologists than by biologists.
Psychologists have tended to take an anthropocentric approach, seeking to understand humanlike performance in other species and perhaps interpreting their findings by reference to an assumed phylogenetic scale.
Anthropocentrism is not the same as anthropomorphism, or interpreting animal behavior as if it was caused by humanlike thought processes. Explicit anthropomorphism is thought to have been rejected with the adoption of Morgan’s Canon early in the 1900s, but cannot be done away with entirely.
The ecological or biological approach to cognition consists of analyzing the kinds of information processing
animals do in situations of ecological importance like foraging, choosing mates, finding their way around. With this approach, species are compared with reference
to evolutionary and ecological relationships.
After a long history in which comparative psychology developed largely independently of biological studies of behavior, contemporary research on comparative cognition is increasingly integrating these two approaches and making rich connections with other biological sciences.

Answer 4

1.3.1 The anthropocentric approach

Comparative psychology began with Darwin’s claim—profoundly shocking at the time—that humans are similar to other species in mental as well as physical characteristics. Darwin (1871) claimed that other animals differ cognitively from humans in degree but not in kind. That is to say, animals share human abilities such as reasoning, memory, language, and aesthetic sensibility, but generally they possess them to a lesser degree.

His emphasis was on continuity among species rather than diversity, the other side of the evolutionary
coin (Rozin and Schull 1988). Acceptance of continuity has led to using animals in psychology as little furry or feathery people, model systems for studying general
processes of learning, memory, decision-making, even psychopathology and their neural and genetic underpinnings. Thus this approach can be characterized as anthropocentric because it is concerned primarily with issues related to human psychology.

Psychologists studying animal cognition, in contrast, used behavior as a window onto processes of memory and representation (Wasserman 1984). Initially, much of
their research used learned behavior of rats and pigeons in the laboratory to analyze processes that were being successfully studied in people, such as memory for lists of items, concept formation, and attention (cf., Hulse, Fowler, and Honig 1978).
Research on animal cognition based on the anthropocentric approach has three important characteristics.
- First, it focuses on memory, representation, and other kinds of information processing that can be identified in people.
- Second, such research is implicitly comparative, in that other species like parrots or pigeons are
compared with humans, but the choice of species is often based more on convenience than on evolutionary considerations.
- Finally, traditional discussions of anthropocentric research were pervaded by the incorrect and misleading notion of a phylogenetic scale or scala naturae (Hodos and Campbell 1969).

This is the idea that evolution is a continuous ladder of improvement, from ‘‘lowly’’ worms and slugs,
through fish, amphibians and reptiles, to birds and mammals. Humans, needless to say, are at the pinnacle of evolution in this scheme. But present-day species cannot be lined up in this way. People are not more highly evolved fish, birds, rats, or even chimpanzees. Correct inferences about the relationship between cognitive or brain processes in humans and those in nonhumans depend on a detailed appreciation
of the biology of ‘‘animal models’’ (Preuss 1995 ; Papini 2008).

Nevertheless, studying a few very diverse species, may be the best way to reveal processes general to all species (Bitterman 2000; Papini 2002). Exactly this
approach to genome mapping has provided stunning support for generality: species as diverse as fruitflies, mice and humans are turning out to share unexpected numbers of genes and basic developmental processes (see Robinson 2004; Papini 2008). In addition, the rigorous methodology and the principles developed with traditional psychological studies of animals are essential to more biologically focused research (e.g., Timberlake, Schaal, and Steinmetz 2005).

1.3.2 Anthropocentrism, anthropomorphism, and Morgan’s Canon

Darwin’s supporters, primary among them George Romanes (1892) set out to collect anecdotes appearing to prove animals could think and solve problems the way people do. Their approach was not just anthropocentric and anthropomorphic, explaining animals’ apparently clever problem solving in terms of human-like thinking and reasoning. But as we have seen in the case of the nutcracking crows, just because an animal’s behavior looks to the casual observer does not mean it can be explained in the same way.

C. Lloyd Morgan also observed animals in a systematic way but is now best known for stating a principle commonly taken as forbidding unsupported anthropomorphism. What is the ‘‘psychological scale’’? In contemporary practice ‘‘lower’’ usually means associative learning, that is, classical and instrumental conditioning or untrained species-specific responses. ‘‘Higher’’ is reasoning, planning, insight, in short any cognitive process other than associative learning.

For an example, suppose that crows had been found to drop nuts in front of cars more than on the empty road. An obvious ‘‘simple’’ explanation is that they had been
reinforced more often when dropping a nut when a car was coming than when the road was empty and thereby had learned to discriminate these two situations. A ‘‘higher,’’ anthropomorphic, explanation might be that having seen fallen nuts crushed by cars the insightful crows reasoned that they could drop the nuts themselves. Morgan apparently thought (Sober 2005), the simplest account of any behavior is arguably the anthropomorphic one, that behavior analogous to ours is the product of a similar cognitive process.

A reasonable modern interpretation of the Canon (Sober 2005) is that a bias in favor of simple associative explanations is justified because basic conditioning mechanisms are widespread in the animal kingdom, in which they have been sought (Papini 2008). Thus they may be evolutionarily very old, present in species ancestral to all present-day animals and reflecting adaptations to universal causal regularities in the world and/or fundamental properties of neural circuits.

At the theoretical level, such experience in Pavlovian (stimulus-stimulus) or instrumental (response-stimulus) conditioning has traditionally been thought of as strengthening excitatory or inhibitory connections
between event representations. Thus one might say that any cognitive performance that does not result from experience of contingencies between events and/or cannot be explained in terms of excitatory and/or inhibitory connections is nonassociative.

Path integration is one example: an animal moving in a winding path from home implicitly integrates distance and direction information into a vector leading straight home. As another, on one view of conditioning the
flow of events in time is encoded as such and computed on to compare rates of food presentation during a signal and in its absence. Other nonassociative cognitive processes which might be (but rarely if ever have been) demonstrated in nonhuman include imitation, that is, storing a representation of an actor’s behavior and later
reproducing the behavior; insight; and any kind of reasoning or higher-order representations or computations on event representations.

discriminating nonassociative ‘‘higher’’ processes from associative ones is seldom straightforward, in part because the learning resulting from associative procedures may have subtle and interesting cognitive content. In any case, the goal of comparative research should be understanding the cognitive mechanisms underlying animal behavior in their full variety and complexity rather than partitioning them into rational or nonassociative vs. associative (Papineau and Heyes 2006).

In conclusion, neither blanket anthropomorphism nor complete anthropodenial is the answer (Mitchell 2005). Evolutionary continuity justifies anthropomorphism as a
source of hypotheses. When it comes to comparing human cognition with that of other species, it is most likely that—just as with our genes and other physical
characters—we will find some processes shared with many other species, some with only a few, and some that are uniquely human. One of the most exciting aspects of contemporary research on comparative cognition is the increasing detail and subtlety in our picture of how other species’ minds are both like and not like ours.

Question 5

2. Evolution, Behavior, and Cognition: A Primer

2. 1. Testing adaptation

Answer 5

‘‘Drab coloration is an adaptation for reducing detection by visual predators.’’ ‘‘Bats’ sonar is an adaptation for detecting flying insects in the dark.’’ ‘‘Reasoning ability is an adaptation to conditions in early hunter-gatherer societies.’’ To say that some characteristic of an animal’s structure, behavior, or cognition is an adaptation is to assert that it has evolved through natural selection. But selection has occurred in the past, so how can we ever test such a statement?

Aren’t hypotheses about adaptation no better than Kipling’s Just-So Stories(Gould and Lewontin 1979) like ‘‘The Elephant’s Child,’’ which explains that elephants have long trunks because a hungry crocodile once
stretched the nose of a curious young elephant? Perhaps just-so stories can be concocted for most situations, but in fact serious ideas about adaptation are testable using direct observation and experiment, model building, or the comparative method. In the best
possible cases, all three approaches can be used in a complementary way.

2.1.1 Testing present function

A character can serve a function in the present without having been selected for that function, that is, without being an adaptation for it. Function may change over
evolutionary time (Williams 1966). To take a nonevolutionary analogy, in big cities like Rome or New York one sometimes sees groups of tourists all wearing identical hats. Hats are designed (adapted) to protect the head. Originally, tour organizers may have found it convenient to give out souvenir hats that were all the same, but that having happened, the hats now serve the function of allowing members of a group to identify one another and stick together.

Evolution and present-day function are not unrelated, however. Demonstrations that a behavior serves a particular function increase confidence in the hypothesis that that function has contributed to its evolution (Cuthill 2005). A classic example of an
experimental test of current function comes from Tinbergen’s (Tinbergen et al. 1963) studies of eggshell removal in gulls. Soon after their eggs hatch, black-headed gulls (Larus ridibundus) pick up the empty eggshells, fly off, and drop them some way from
the nest. Why should a bird leave its vulnerable chicks for even a few seconds to engage in this behavior?

Maybe the white insides of broken shells attract predators. To test this hypothesis, Tinbergen and his colleagues distributed single gull eggs around the
dunes where the black-headed gulls nest. Some of these decoy eggs had broken eggshells placed nearby; others were isolated. The eggs near broken shells disappeared sooner, eaten by crows and herring gulls, than the less conspicuous isolated eggs. Thus removing broken eggshells from the nest functions to protect offspring from the predators found where the gulls nest.

This suggests a comparative hypothesis: gull species nesting in areas without this same predation pressure should not remove empty shells from their nests. Kittiwakes (Rissa tridactyla) provide a natural test of this
hypothesis. These gulls nest on small ledges on steep cliffs, inaccessible to most predators. Kittiwakes’ behavior differs from that of ground-nesting gull species in several ways that can be seen as adaptations to nesting on cliffs (Cullen 1957). Among other things, they do not remove broken eggshells from their nests.

2.1.2 Adaptation as design

Many features of animals’ structure and behavior seem so perfectly suited to their function that they seem unlikely to have arisen by chance. The eyes of vertebrates, the sonar of bats, the nest building and parental behavior of birds: all seem designed
to accomplish their ends. Often, designs in biology are remarkably like what engineers would build to achieve the same goals. These considerations seem to
compel the conclusion that intricate structures and behaviors like eyes, ears, and eggshell removal must be evolved adaptations.

In pre-Darwinian days, however, such arguments from design were used as evidence for a divine creator (see Dennett 1995). Darwin’s genius lay in deducing how natural causes produce the same end. A major contribution of behavioral ecology has been the use of formal optimality models to study adaptation. Working out the optimal behavior for a given situation is a way of specifying the best design. One beauty of precise optimality arguments is that in principle they can be shown to be false.

For example, the schooling behavior of fish had been thought to save energy for each individual by allowing it to swim in the eddies from its neighbors. However, detailed consideration of the hydrodynamics of swimming fish shows that in fact individuals of some species do not position themselves in so as to benefit as much as they could from the way the water is moved by other fish in the school. Thus, although hydrodynamic advantage may have contributed to the evolution of schooling behavior, other selective forces
must have been involved (see Dawkins 1995).

This is an example of how a model’s predictions can fail because the modeler failed to take into account all the relevant factors. Such failures may lead to more complex models incorporating tradeoffs among competing selection pressures. In any case, evolution has not necessarily always produced the absolute optimum. Among other reasons, selection can work
only on preexisting variations among individuals, including the variations thrown up by random mutations.

2.1.3 The comparative method

At most, experimental tests of function or observations of natural selection in action like the studies of Galapagos finches can be done on only a few species. For a look at the broad sweep of evolution, at whether an important selection pressure has produced similar patterns across many species, the comparative method is essential. We have already met an informal example in the study of eggshell removal by gulls nesting in different habitats.

In general, a comparative test of the adaptive value of a character consists of obtaining data from a large number of species and relating the degree to which they display the character with the degree to which the hypothesized selection pressure is present (Harvey and Pagel 1991; for introduction see Sherry 2006). It must be applied together with good information about evolutionary relationships (i.e., phylogeny) so similarity due to common selection pressures can be distinguished from similarity due to descent from common ancestors.

Conclusions about adaptation may therefore change with changes in the amount and quality of information used to construct the associated phylogeny. Animals live in all sorts of places and in an amazing variety of kinds of social groups. Some are solitary and cryptic except during mating. Others, like the wildebeest of the African plains or the caribou of the American Arctic, form enormous herds. Breeding may take place promiscuously, or between members of monogamous
pairs or, among other possibilities, in a polygynous mating system, in which a few males may each control access to a harem of many females.

Why have all these different social arrangements evolved? One approach to answering this question is
to see if social structure can be related to ecology. Vulnerability to predators, what food a species eats and its spatial and seasonal distribution, the availability of nesting sites—all these and other variables can be related to social organization in a variety of animal groups. For example, in African ungulates, body size, habitat, group size, and mating system are related in the way (Jarman 1974).

Smaller species need high-quality food because they have a high metabolic rate. They primarily seek fruits and buds in the forest. Because these foods are relatively sparse, the animals cannot form large groups, and there is no opportunity for one male to monopolize many females. Rather, the small-bodied forest species are found alone or in pairs. The large-bodied species graze relatively unselectively on the plains, on food
that is locally very abundant but which varies seasonally in distribution with rainfall.

Thus species like wildebeest tend to form large herds that migrate long distances with the seasons. Being in a group opens the opportunity for one male to monopolize several females. Hence, polygyny rather than monogamy tends to be found in the large grassland species. By itself, especially as summarized in a paragraph, this account seems like a Just-So
Story. Several things make it much more than that. For one, a similar account can be given of social structure in other animal groups, including birds and primates (Cuthill 2005; Danchin, Giraldeau, and Cezilly 2008).

This is what would be expected if social structure is the outcome of fundamental selection pressures like the distributions of food and predators and not just associated with ecology in ungulates by chance. For
another, more detailed comparative analyses have tended to uphold the conclusions from categorical analyses. Consider one correlate of social structure, sexual dimorphism in body size, that is, the degree to which males and females are different sizes (Clutton-Brock and Harvey 1984).

In a variety of animal groups, males tend to be about the same size as females in species that form breeding
pairs, whereas males tend to be larger than females in polygynous species. One possible explanation of this relationship is that large males have an advantage in defending females from rival males. Among primates, polygynous species may live in one-male or multi-male groups. Each male dominates more females in one-male groups. Sexual dimorphism in primates, measured as ratio of male weight to female weight, is related
to mating system just as this discussion predicts.

Results like those shown in Figure 2.2 and Table 2.1 must not be distorted by unequal degrees of relatedness among the species being considered. If the species within each ecological category are more closely related to each other than they are to species in other ecological categories, differences among categories could reflect descent from a common ancestor rather than common selection pressures. One way to deal with this problem is to look at different groups of species.

For instance, the same relationship between sexual dimorphism and breeding system is found in several independently evolved animal groups, suggesting that it is indeed related to the degree to which males compete for females. Although Figure 2.2 shows a significant positive relationship between sexual
dimorphism in body size and number of females per male in the breeding group, the error bars indicate that considerable variation is still unaccounted for.

Correlations between characters and ecology across large numbers of species almost always use data from many sources, and inevitably some data points will represent larger numbers of more careful observations than others. However, if enough species are sampled, random errors should balance each other out and genuine relationships reveal themselves. The variables examined also need to be good measures of the
ecological factors being considered.

For instance, ratio of females to males in the
breeding group might not be the best measure of intermale competition, the factor hypothesized to favor large-bodied males, and body size is probably influenced by factors other than social structure, such as whether the animals live in trees or on the ground (Clutton-Brock and Harvey 1977). Obvious exceptions to an overall relationship can be instructive. Figure 2.3 shows an example based on the allometric relationship among the sizes of body parts. Allometry refers to the principle that animals with bigger bodies have, on average, bigger body parts.

A plot of the size of any structure against total body size has a characteristic slope, with most points clustered close to the overall regression line. The volume of the hippocampus, a brain structure important for memory, particularly spatial memory, in mammals and birds is plotted against body weight and against volume of the telencephalon (most of the rest of the brain) for a large number of genera of European birds.

Three points stand out as being substantially
above the overall regression lines indicating that three groups of birds have larger hippocampi than expected for their body and brain sizes. These all contain species
that store food for the winter and retrieve it using long-lasting spatial memory. food storing evolved together with a relatively large hippocampus.

Question 6

2. Evolution, Behavior, and Cognition: A Primer

2. 2. Mapping phylogeny

Answer 6

Correlation is not evolutionary causation. The association between food storing and a relatively large hippocampus does not tell us about the sequence of events in evolution. Maybe food-storing species evolved an unusually large hippocampus for some unknown reason and it then allowed them to benefit from storing food. Or maybe rather than ask why some birds have such a large hippocampus relative to brain and body size we should be asking why other birds have such a small one (Deacon 1995).

Such questions have to do with what ancestral species were like and how and why they changed, but suggestions of the answers to them can be found by looking at present-day species, given some reasonable assumptions about how evolution works. This is the study of phylogeny, or the reconstruction of the tree of life, the branching relationships among species during
evolution (Stearns and Hoekstra 2005).

Suppose we have a bat, a bird, and a monkey. The bat is like the bird in having wings, but it is like the monkey in having fur instead of feathers, lactating, and giving
birth to live young instead of laying eggs. On the basis of these four characters, we would classify bats as more closely related to monkeys—that is, having a more recent common ancestor—than to birds because bats and monkeys have more characters in common. Moreover, although bats and birds both have wings, they differ embryologically and in details of structure.

Thus they are homoplasies (or analogies), not homologies, that is, they have evolved from different ancestors and converged on a similar shape due to common selection pressures of an aerial way of life rather than being descended from a common winged ancestor. Differences between the bat’s limbs and the monkey’s reflect a third evolutionary outcome,
divergence from a comparatively recent common (mammalian) ancestor (for further discussion see Papini 2002; Papini 2008).

Biological classification is hierarchical. Figure 2.4 shows three ways of representing the nested relationships among species. A phylogenetic tree represents the divergence among species over time. The times at which species diverged from an ancestral state can be tied down by examining the fossil record and/or from molecular evidence based on species differences in DNA and/or other molecules and assumptions about
the rate of random mutation of DNA.

Figure 2.5 shows the phylogeny of primates based on molecular evidence. Many diagrams of primate phylogenies betray our continuing belief that humans are the ‘‘highest’’ primates by putting them at the top
(Hodos and Campbell 1969; Nee 2005). In fact, the arrangement of species branching from a particular node is largely arbitrary. What matters is the nodes (i.e., connections), not which ones are higher on the page or further to the right or left. Figure 2.5 puts chimpanzees at the top to emphasize the sequence in which the species diverged from common ancestors.

The classification of organisms into clades, or groups descended from a common ancestor, can be based on characters of present-day species alone. Nowadays
an important part of this process is comparison of gene sequences and proteins and use of sophisticated statistical techniques that take into account large numbers of characters (see Pagel 1999).

But the simple example in Table 2.2 is enough to show the logic of phylogenetic reconstruction. Without knowing anything about genes or the fossil record, we could infer from the table that bats and monkeys share an ancestor that had fur, gave birth to live young and lactated (i.e., a mammalian ancestor) that was not ancestral to birds. Such inferences rely on the notion that any particular evolutionary change is improbable.

For a new species to evolve, an advantageous rather than a deleterious or lethal mutation has to occur and spread. It is therefore more likely that shared characteristics were present in a common ancestor than that they evolved several times independently. Representations of cladistic classification can display the characters that have changed as species diverged, as in Figure 2.6. Finally, although the classification of organisms into species, genera, families, and so on is also hierarchical, traditional classifications of species groups do not always correspond so closely to the other classifications as in Figure 2.4c.

Question 7

2. Evolution, Behavior, and Cognition: A Primer

2. 3. Evolution, cognition, and the structure of behavior

Answer 7

2.3.1 Behavior systems

Behavior is organized into functional systems like hunger, fear, and sexual behavior, called instincts by Tinbergen and other classical ethologists. These are hierarchical organizations of motor patterns that share some proximate causal factors (Timberlake 1994; Hogan 1994b). For example, an animal’s hunger system includes the behavior patterns that change in frequency, intensity, or probability when it has been deprived of food and/or is in the presence of food.

For a chicken, these might be walking around, scratching the ground, and pecking. A behavior system also includes relevant stimulus processing (perceptual) mechanisms and central mechanisms that coordinate external and internal inputs. In the case of the hunger system in a chicken, a central motivational mechanism integrates the bird’s state of depletion or satiation with visual information to determine whether or not it will peck at what it sees (Hogan 1994b).

Cognitive mechanisms are part of this organization, too.
Whether the chicken pecks at the thing in front of it may be influenced by what it is attending to and by past learning about the consequences of pecking. As just described, behavior systems are defined causally (Hogan 1994b), in terms of internal and external causal factors rather than immediate outcome or apparent
goal. However, the causal organization of behavior must make functional sense.

An animal that ignored food while starving or approached predators rather than hiding or running away would be unlikely to have as many offspring as one that ate when hungry and fled from danger. Animals that ignore food when deprived or behave in a
friendly manner toward predators have been weeded out by natural selection not because they are ‘‘too stupid’’ to forsee the dire consequences of their acts but because they leave fewer copies of their genes than do individuals whose motivational and cognitive mechanisms result—blindly—in their being better-nourished and less preyed upon.

This relationship is depicted in Figure 2.8. Natural selection shapes cognition in an indirect way. Cognition—processing environmental information— results in behavior. That behavior has an immediate consequence such as ingesting food, depositing sperm in a fertile female, strengthening a nest. In the long run, such consequences have a measurable impact on the individual’s fitness and thereby on the representation of genes contributing to development of the mechanisms that generate that behavior.

With few exceptions, like nest-building and burrowing, behavior does not leave fossils. But the evolution of behavior can nevertheless be inferred from phylogeny,
as indicated in Figure 2.6. In terms of the organization of behavior systems shown in Figure 2.7, species differences could evolve in sensory, motor, or central mechanisms. For instance, the range of energies detectable by the senses could expand or contract, new motor patterns could appear, and/or the central coordination of input and output could change.

The evolution of behavior can be traced at a more
detailed level, too. For instance, species differences in motor patterns may be analyzed into differences in muscular and skeletal anatomy and patterns of firing
in nerve cells (Lauder and Reilly 1996). Species differences in visual sensitivity related to differences in the kind of light prevalent in different environments might be related to differences in photopigments and the genes for producing them.

The loss of bat-avoidance behavior by moths on Tahiti is an example of evolutionary change nicely accommodated by this way of thinking about behavior.
The raison d’eˆ tre for hearing in most moths is to avoid bats, which search for moths in the dark using ultrasonic cries. Accordingly, a moth’s simple auditory
system is tuned to ultrasonic frequencies because moths can avoid bats by dropping immediately to the ground when they hear one.

Although bats have apparently never been present on the Pacific island of Tahiti, the auditory nerves of the moth species that arrived on Tahiti millions of years ago (endemic species) still fire to bat cries. Nevertheless, when bat cries were played to endemics in flight, they did not drop to the ground like individuals of more recently arrived species. Assuming that the endemics are still capable of altering their flight in response to other stimuli, this pattern of findings indicates that in the absence of selection the sensory input has been decoupled from the motor avoidance response (Fullard, Ratcliffe, and Soutar 2004).

Many morphological (i.e., structural) differences among species result from relatively small changes in developmental programs, that is, from changes in when
specific genes are turned on and off (see Stearns and Hoekstra 2005). A speeding up or slowing down of growth in one part relative to others can result in dramatic changes in shape. The brains of food-storing birds provide one example related to cognition. 

In baby marsh tits (food-storers) and baby blue tits (nonstorers), the whole brain grows rapidly in the first few weeks after hatching. At this stage, the hippocampus develops relative to the rest of the brain in the same way in both species. By around 6 weeks after hatching, when the babies are feeding themselves and the marsh tits are starting to store food, brain growth has slowed down.

However, the marsh tits’ hippocampus continues to grow, so that the typical foodstorers’ larger hippocampal size relative to the rest of the brain appears by the time memory for storage sites is needed (Healy, Clayton, and Krebs 1994). Magpies (foodstoring corvides and jackdaws (nonstoring corvids) show the same pattern (Healy and Krebs 1993). In the case of marsh tits, experience using spatial memory also contributes to the species difference in hippocampus, but blue tits are not influenced by experience in the same way as marsh tits (Clayton 1995).

Darwin was deeply impressed by how behavior as well as structure could be artificially selected by animal breeders. And in The Origin of Species he speculated on how complex and intricate behaviors like the comb-building behaviors of honey bees could have evolved in small steps. Nowadays, genetic engineering can be used to demonstrate that particular genes contribute to particular behaviors or cognitive processes and to analyze the mechanisms by which they do so (Mayford, Abel, and Kandel 1995; Fitzpatrick et al. 2005).

Natural selection can provide molecular geneticists with opportunities to dissect how genetic changes have produced species differences, including differences in cognition and behavior. Bringing together information derived from genetic engineering with phylogenies of real species offers exciting possibilities for research on the mechanisms of evolutionary change (see Fitzpatrick et al. 2005; Grant and Grant 2008).

Most tests of adaptation mentioned in Section 2.1 involve comparing different species or groups of species: ground-nesting vs. cliff-nesting gulls, solitary vs. social species of ungulates, and so on. Naturally, any such comparison must be done carefully. For example, when correlating social group size and male body size, it is important to be sure the values going into the analysis are representative of the species and to take
account of other variables that might be confounded with the variables of interest.

Comparing cognition across species encounters its own particular difficulties, which largely arise from the fact that behavior is influenced by a host of processes that are not specifically cognitive. As a result, conclusions like ‘‘species A has more of ability X than species B’’ always need to be viewed critically. The same is true in comparisons of genetically manipulated animals. This section introduces some of the general problems in doing comparative research on cognition, taking as an example research on male-female differences in spatial memory in different species of rodents.

This is not to imply that such problems have not been addressed in this area; as we will see, many have been dealt with rather well. In many monogamous animals, the male and female occupy a territory together,
whereas in some polygynous species females have relatively small territories where they rear their young, while males range over larger areas, visiting several different females for mating.

These observations suggest that in monogamous species males and females need similar abilities to find their way around and remember the locations of resources in the pair’s territory, whereas in polygynous species males need a better-developed ability to process and remember spatial information than do
females. This hypothesis about the relationship between spatial cognition and mating system has inspired research on sex differences in brain and spatial cognition in several groups of rodents (Gaulin 1995; Jacobs 1995; Sherry 2006).

It is arguably the most coherent and best-supported of several proposed evolutionary explanations for the sex differences in spatial cognition observed in a variety of mammals, including humans (see Jones, Braithwaite, and Healy 2003). The specific hypothesis here is that males and females do not differ in spatial ability in monogamous species whereas there is a difference in favor of males in polygynous species. But to evaluate it, we cannot necessarily just test males and females of a number of different species all in the same way because a test standardized in terms of physical variables may affect different species differently.

For instance, animals that become frightened and stay close to the walls in a big open space might take longer than bolder animals to learn to swim straight to the
dry platform in the middle of a pool of water. Recent research on animal personality has provided plenty of evidence for stable within- and betweenspecies differences in behavior that could influence the outcome of cognitive tests as this example suggests. If the animals are rewarded with food, we need to be sure all species are equally hungry and equally fond of the reward provided.

If we compare them on discrimination learning, we need to know that they process the stimuli involved in the same way, that is, we need to know something about their sensory systems. Such considerations underline the importance of what Macphail (1982,
1987) called contextual variables. Within any species, many aspects of the experimental context, some much less obvious than timidity or reward size, can affect
what animals do. Therefore, any species difference on a single task could reflect different effects of contextual variables on performance rather than the cognitive
ability that performance is supposed to measure.

One proposed solution to this problem is systematic variation (Bitterman 1965). This means testing the animals under several values of relevant contextual variables. For instance, the difficulty of the task should be varied over a wide range. Gaulin and Fitzgerald (1989) did just that by using seven different mazes to compare spatial learning in monogamous prairie voles (Microtus ochorogaster) and polygynous meadow voles (M. pennsylvanicus). Meadow vole males performed better than meadow vole females on all the mazes, but, as predicted, there was no sex difference in the prairie voles.

Importantly, the mazes seem to be a fair test of species differences in that both species score about the same on average on any given maze. They are also not so hard that most animals fail or so easy that everyone does perfectly, which is important because ‘‘floor’’ or ‘‘ceiling’’ effects, respectively, could obscure group differences. Bitterman (1965), one species fails to show some effect shown by another or shows it to a much smaller degree. Clearly, if it is already known that the strength of this effect in species that show it is influenced by some contextual variable, then this same variable should be manipulated with the second species to be sure it was not just at an unfavorable level initially.

Here, systematic variation amounts to trying to reject the null hypothesis that no factor other than differences in cognition is responsible for differences in performance (Kamil 1988). To return to our case study, it might be suggested that sex differences in activity are
responsible for sex differences in performance in spatial tasks. This possibility has been rendered implausible by showing that males’ and females’ activity levels are similar under a range of conditions (Gaulin, FitzGerald, and Wartell 1990).

But a skeptic might then suggest another confounding factor, further systematic variation would have to be done, and so on ad infinitum. Kamil’s proposed solution to this problem is, instead of systematically varying factors within a given task, to vary the tasks. For instance, if food-storing and nonstoring species differ in ability to process and remember spatial information, these differences ought to be detectable in a variety of different spatial tasks. There may of course be tasks
or species for which contextual variables are overwhelmingly important, but if enough tasks are used, the results should converge on a single conclusion.

Kamil and his colleagues have used this approach with considerable success to compare memory for spatial information in food-storing vs. nonstoring species of birds. The other side of systematic variation is emphasized by Papini (2008): if an independent variable affects species in the same way, even if their levels of performance generally differ quantitatively, this is evidence for a shared process. Figure 2.10 provides
an example. Although male meadow voles perform better than females, their errors still increase with maze complexity.

Systematic variation appears frequently throughout this
book as a way to discover whether very different species, exhibiting behaviors as different as speaking vs. pressing a key vs. digging up a worm, have access to the same kinds of cognitive processes. The tests of blindsight in monkeys described in Box 1.1 are
an example. This approach is also, referred to as testing for functional similarity. Most importantly, examples of what can be learned from systematic variation underline the principle that conclusions about species differences in cognition must always be based on more than a single test.

Ideally, a thorough comparative test of an ecological hypothesis includes tests on which the species are predicted not to differ, or—even better—to differ in the opposite direction. Such tests can help to rule out the possibility that one group performs better than another because of some general factor like how well they adjust to the lab.

In food-storing species of corvids (the crow family, including jays and nutcrackers; Box 1.4), some species are highly social while others are not. Therefore, the pattern of species differences in social cognition may differ from that in spatial cognition (Balda and Kamil 2006). Sex differences in spatial behavior related to space use in the wild may be present only in the breeding season (Galea et al. 1994; Sherry 2006). Such
seasonal or developmental changes within individuals of the same species offer excellent opportunities for testing adaptive relationships among cognition, brain,
and natural behavior with minimal confounds from contextual variables.

An example is the comparison of spatial and other kinds of memory in white-footed mice exposed
to summerlike vs. winterlike photoperiods (Pyter, Reader, and Nelson 2005). However, even comparisons within a species may be subject to motivational or other
confounds. For example, the time available for feeding may differ when animals live in days of different lengths, and/or the animals in short days may reduce their activity or metabolic rate.

A general problem with applying the comparative method to behavior and cognition is getting enough independent comparisons. One solution to the practical difficulties of testing large numbers of species is to build up a sample gradually by comparing two species at a time, in this case one monogamous species with one closely related polygynous species, but we need to be able to find a sufficient number of lineages in which monogamy arose separately. Research relating spatial ability to mating systems has been done on, among other rodents, voles (Microtus) and mice (Peromyscus), and of course the hypothesis could also be tested on birds with appropriate mating systems (Jones, Braithwaite, and Healy 2003).

Exceptional spatial ability may be associated with other exceptional demands on spatial learning and/or memory in the wild. For instance, birds that migrate might be expected to use memory and spatial learning more than relatively sedentary populations, not necessarily because they actually need learning to migrate, but because they need to acquire spatial and other information about each of the places where they
spend a few months at the ends of their travels, and perhaps at stopovers along the way.

They might also form long-term memories for the areas where they regularly spend part of the year, so as not to waste time relearning their stable features. There is some evidence consistent with this hypothesis (e.g., Cristol et al. 2003; MettkeHofmann and Gwinner 2003). Not only amount but kind of spatial learning might be expected to be associated with different ecological demands. For example, individuals living in different kinds of habitats might rely on different kinds of spatial cues. In one test of this notion, Odling-Smee and Braithwaite (2003) found that stickleback fish from ponds relied more on landmarks than fish of the same species from fastmoving streams.

Question 8

2. Evolution, Behavior, and Cognition: A Primer

2. 4. Evolution and the brain

Answer 8

2.4.1 Patterns of vertebrate brain evolution

To look for patterns in a large sample of species, it is a lot easier to measure brains than to measure behavior and infer cognitive structures. As a result, compared to
what we know about the distribution of any cognitive ability across the animal kingdom, we know vastly more about the brain, at least in vertebrates. Figure 2.11
shows the relationship of brain weight to body weight in the major groups of vertebrates. The polygons enclosing data from each taxonomic group (taxon) indicate that brain size can vary considerably even for animals of a given group with a given body weight, as illustrated for mammals in the lower panel of Figure 2.11.

There is a trend for larger brains during vertebrate evolution. For instance, birds are thought to have evolved from a primitive reptile, and the polygon for birds is entirely above that for reptiles, indicating that in general birds have larger brains than reptiles
of equivalent size. On the whole, mammals have the largest brains for their body weight, but small mammals overlap considerably with birds. Within mammals,
humans are the species farthest in perpendicular distance above the group regression line (details of each taxon in Chapter 4 of Striedter 2005).

Within a lineage, why do some species have larger brains relative to their body weights than others? Brains are metabolically costly to maintain (Laughlin 2001), so
there must be some advantage to having a large brain. Not surprisingly, hypotheses about the function of relatively large brains have focused on the assumed connection between brains and cognitive abilities. For instance, the ‘‘foraging intelligence hypothesis’’ of primate brain size proposes that fruit-eating species need excellent spatial and temporal learning abilities for tracking the locations and ripeness of items that are scattered widely throughout the forest whereas leaf eaters do not need such abilities.

The ‘‘social intelligence hypothesis’’ (Chapter 12) suggests that animals living in large groups in which individuals have differentiated and ever-changing
social roles need to keep track of the identities of large numbers of individuals and their interactions. Tests of the various versions of these hypotheses have relied on
comparative studies relating primates’ brain size to proxies for cognitive abilities such as type of foraging niche or social group size (review in van Schaik and Deaner 2003; Healy and Rowe 2007).

Among birds, parrots and corvids have the biggest brains for their body sizes. As we will see, some corvids may have social and tool-using abilities comparable to those of some primates. These, along with relatively large brains, appear to represent convergent evolution in separate vertebrate lineages (Emery and Clayton 2004). Relatively new are the comparative studies of primates and birds described in Box 2.2 indicating that brain size is related to propensity for innovation. To the extent that foraging on ephemeral food sources, managing social relations, and acquiring novel behaviors call on common abilities, these explanations for the evolution of large brains need not be mutually exclusive (Striedter 2005).

In any case, most accounts of relative brain size in terms of complex behaviors are still largely speculative pending more direct evidence about the neural substrates of the behaviors in question (Healy
and Rowe 2007). The foregoing discussion addresses the whole brain, but the relationship of relative hippocampus size to food-storing in birds depicted in Figure 2.3 suggests that maybe we should be looking at how individual parts of the brain evolve in
association with specific behaviors or ecological variables.

Whether brain evolution is concerted or mosaic, that is, whether brain size evolves as a whole or through selection on particular parts, is a contentious question in comparative neuroanatomy (see discussions accompanying Finlay, Darlington, and Nicastro 2001 and Striedter 2006). Figure 2.11 is consistent with concerted evolution because it shows an evolutionary trend toward larger brains. Historically, brain evolution was thought to be a matter of adding new, more
advanced, structures to primitive ones in a linear fashion leading up to primates and humans.

Hence the prevalence of prefixes such as ‘‘paleo,’’ ‘‘neo,’’ and ‘‘archeo’’ to label structures in traditional brain anatomy. It is now recognized that all vertebrate brains have the same basic parts, although their relative sizes and detailed structures are characteristic of each vertebrate group (Avian Brain Nomenclature Consortium 2005). Within a lineage, larger brains are not just scaled up versions of smaller ones. Bigger brains need a more modular organization (Box 2.3), and this well might lead to cognitive differences between bigand small-brained species within a group, for example, primates vs. rodents or parrots vs. canaries.

The proportion of the brain occupied by particular structures such as the neocortex also tends to differ in a systematic way in largerbrained species, apparently consistent with mosaic evolution. However, on one
theory (Finlay, Darlington, and Nicastro 2001) most of this variation is consistent with concerted evolution because it reflects the way in which common
processes of very early brain development produce larger brains.

Indeed, a recent survey (Striedter 2005) finds that the majority of the evidence is consistent with concerted evolution in that within a given taxon, and after taking into account developmental constraints, the relative size of a given structure generally does not show very dramatic deviations across species. ‘‘Not very dramatic’’
means not more that about a 2- or 3-fold difference in size relative to the rest of the brain.

Within this context, the hippocampus–food storing story is ‘‘wonderful’’ (Striedter 2005, 173) as a potential example of at least mildly mosaic evolution. It is also an instructive case study of the challenges of trying to
connect brain, behavior, and cognition in a rigorous way. 2.4.2 Hippocampus and food storing in birds
The principle of proper mass (Jerison 1973) as a tenet of comparative neuroanatomy says that the more important a function is for a species, the more brain area will be devoted to it.

This principle is most sensibly interpreted as applying to the size of a structure relative to other parts of the brain in comparisons of reasonably close relatives (Striedter 2005). Sensory and motor areas provide some spectacular illustrations. For instance, the superior colliculus, a visual processing area, is nine
times larger in a 13-lined ground squirrel (a diurnal species) than in a laboratory rat (nocturnal), and in the blind mole rat, which spends its life underground, it is 38 times smaller than in a hamster.

In the very dextrous raccoon, the sensory and motor areas devoted to the paws are greatly enlarged compared to those in other nonprimates (see Streidter 2005). Although these examples are exceptional in
quantitative terms, because sensory systems are clearly evolved to allow each species to discriminate the stimuli most important to it (Chapter 3) it is not
surprising to find sensory specializations reflected in species-specific tweakings of sensory organs and associated brain areas.

However, suggestions that an analogous principle applies to cognition and the brain—in particular to an association between demand for spatial memory in the wild and size of the hippocampus— have been surprisingly controversial (Bolhuis and Macphail 2001; Macphail and Bolhuis 2001; Bolhuis 2005). Cognition is surely not exempt from evolutionary processes, so why should this be? Figure 2.3 shows that among North American families of birds the three families with food-storing species all have, on average, larger hippocampi than expected for the size of the rest of their brains.

The relationship between food storing and performance in tests of spatial memory is discussed in Chapter 7; here we delve into the relationship between food storing and hippocampus suggested by Figure
2.3. One can ask a number of questions about it. For example, what exactly does a bigger relative hippocampus consist of in neuroanatomical terms? How does a comparatively large hippocampus impact on the rest of the brain? How does it improve ability to retrieve stored food? For instance does a relatively large hippocampus increase the capacity or the durability of memory?

These questions are still largely unanswered (see Bolhuis 2005), but some progress has been made in more detailed application of the comparative method to test the basic relationship shown in the figure. This is not to say that the concept of modularity is unproblematical. For example, if we identify a module as a domain-specific kind of information processing, how do we distinguish domains or a ‘‘kinds of information processing’’?

Evolutionary psychologists have promoted the metaphor of the mind as a Swiss Army knife, that is, a general-purpose tool made up entirely of special-purpose devices. But is there a module for everything? If a cheater-detection module (Chapter 12), or a faceprocessing module (Kanwisher 2006), why not hundreds of other modules beside (Fodor 2001;
Buller 2005)? In learning theory the modularity debate takes the form of a debate about adaptive
specializations versus general processes of learning (Section 2.5.2), but forthcoming chapters provide illustrations of how association formation is not the only way of acquiring information.

At the same time, however, many candidate modular learning and memory systems share some fairly general properties such as sensitivity to duration and frequency of events. Thus modularity should not be emphasized at the expense of common features or connectedness. If nothing else, candidate modules are connected by virtue of being contained within the same individual.
Modules may share sensory input systems, and, no matter how specific the triggering information, decision making, and behavioral output of a modular cognitive subsystem, central decision making of some sort is needed to set the animal’s prorities for action.

West-Eberhard (2003) recommends keeping the focus on connectedness and modularity at the same time by
eschewing the term module and referring instead to developmental systems as more or less modular (see also Callebaut and Rasskin-Gutman 2005). Perhaps this recommendation can be applied to cognitive modularity as well. Figure 2.3 classifies birds simply as food-storing or not, but in fact dependence on stored food varies considerably within both parids and corvids. 

For example, the Clark’s nutcracker (Box 1.4) stores one type of food, pinyon pine seeds, very intensely during late summer and depends on its stores throughout the winter. The jackdaw, another corvid, does not store at all, and some other corvids store only moderately.
Similarly, the great tit and blue tit do not store, whereas the willow tit and black capped chickadee store a great deal. These variations suggest looking within families
at hippocampal volume as a function of dependence on storing. This has been done a number of times for both corvids and parids, with results coming out first one way
(e.g., Hampton et al. 1995; Basil et al. 1996) and then another (Brodin and Lundborg 2003) as successive analyses have been more and more refined.

It turns out that, for unexplained reasons, North American corvids and parids tend to have smaller
hippocampi than European species, but when this continent effect is controlled for in cross-species comparisons, relative hippocampus size does correlate with food hoarding status in both corvids and parids (Lucas et al. 2004; Healy, de Kort, and Clayton 2005). Birds that store a lot also tend to have bigger brains overall than expected for their body size, perhaps reflecting sensory or motor specializations in behaviors for storing and retrieving food (Garamszegi and Eens 2004).

These analyses have all assumed that each species fits into a single category of hoarding intensity. However, some food storers such as black-capped chickadees
have a very wide distribution, from rather moderate climates to areas with severe winters. One might expect differences between populations in such species.
Accordingly, when chickadees from Alaska are compared to those from the lowlands of Colorado in tests in the laboratory, the Alaska birds store more, show better spatial but not color memory, and have larger hippocampi relative to brain size (Pravosudov
and Clayton 2002).

Since the birds in this study were taken from the wild, it is not known whether this hippocampal difference is present early in development or results from differences in food hoarding or other experiences in the wild. There are also many unanswered questions about details of hoarding-related changes in the brains
of the chickadees in this and related studies (Bolhuis 2005; Sherry 2006). Research on food-storing birds is but one set of tests of the more general hypothesis
that spatial memory and hippocampus size should be related to demands on spatial memory in the wild (Sherry, Jacobs, and Gaulin 1992).

Much of the work relating spatial learning and memory to territory size and migration discussed in Section 2.3.2 includes studies of the hippocampus (see Sherry 2006). An example involving sex differences comes from cowbirds. The females of several species of cowbirds lay their eggs in other birds’ nests (i.e., they are nest parasites). The females of the brownheaded cowbird (Molothrus ater), a North American species, spend a good deal of their time in the breeding season prospecting for nests where potential hosts are about
to lay.

They need to remember the locations of many nests so as to be able to deposit an egg quickly when the host parent is absent at just the right time in its breeding
cycle. Male brown headed cowbirds share none of this work, so they might be expected to have smaller hippocampi than females. And indeed the predicted sex difference is found for hippocampus relative to the rest of the brain in cowbirds, whereas there is no sex difference in two closely related species that are not nest parasites (Sherry et al. 1993).

Making this story even more interesting, three other
species of cowbirds are found in Argentina, only one of which behaves like the brown headed cowbird. In another, male and female prospect for nests together, and the third is not a nest parasite. Hippocampi of these three species show the pattern of species and sex differences in relative size predicted from the notion that participating in finding host nests requires exceptional memory (Reboreda, Clayton, and Kacelnik
1996). However, unlike the examples involving food storing or territory size, there is as yet little information on spatial memory in any of these birds in standardized
laboratory tests (see Sherry 2006).

Question 9

2. Evolution, Behavior, and Cognition: A Primer

2. 5. What does all this have to do with comparative psychology?

Answer 9

2.5.1 Function and mechanism and the comparative method

The kind of research summarized in the last section was dubbed neuroecology by Bolhuis and Macphail (2001). It has been criticized by these authors (see also Macphail and Bolhuis 2001; Bolhuis 2005) for supposedly confusing answers to Tinbergen’s question about mechanism (e.g., how does cognition or the hippocampus work?) with answers to the question about function (e.g., what is spatial memory or the hippocampus good for?).

This theoretical critique has tended to be combined with a defense of the overwhelming role of general processes in learning and memory and/ or with claims that hippocampus, food-storing, and spatial memory are at best only weakly related. Clearly, correlating features of the brain with food storing or any other ecologically
relevant behavior does not show us directly how the brain works but rather what it allows the animal to do.

Nevertheless, knowing what something does can provide valuable clues as to how it works. Figure2.12 is an example borrowed from Richerson and Boyd (2005).
To quote Sherry (2005, 449), ‘‘Causal explanations must meet design criteria that are set by the function of behavior.’’ Therefore, the study of adaptation (or current function), with which we began this chapter, has a role in the study of cognition and the brain. Acritical application of the comparative method–a solid data set with many cases of independent evolution and checks that other areas of the brain are not also correlated with the same behavior or ecological factor—provides strong evidence that particular behavioral and neural characters evolved together.

Additional data could perhaps give us a picture of the sequence of eventsin evolution. For example, de Kort and Clayton (2006) suggest that a phylogeny of corvids shows ancestral corvids were moderate cachers, and
therefore that food caching has become more intense in some species while being lost in others. And of course the correlational evidence characteristic of the comparative method is rarely interpreted in isolation. For example, behavioral and lesion studies of individual species clearly show that the hippocampus is involved in spatial memory and cache retrieval.

In the example in Box 2.2, we know very little about what innovativeness or behavioral flexibility means in terms of specific cognitive and brain mechanisms, so
this is a case in which findings from the comparative method may suggest new kinds of naturalistic tasks that could be used to compare species behaviorally. The idea that cognitive science can advance by analyzing the information processing tasks that organisms are designed to do has been profitably applied to the study
of perception (Marr 1982; Shepard 1994). 

Among the most prolific and eloquent proponents of the view that thinking about the evolved function of cognition is the best way to understand how it works are the evolutionary psychologists Leda Cosmides and John Tooby (e.g., Cosmides and Tooby 1995; Tooby and Cosmides 1995). One prediction of this adaptationist point of view is that distinguishable cognitive mechanisms or modules (Box 2.3) will evolve whenever the informationprocessing problems a species has to solve require different, functionally incompatible, kinds of computations (Sherry and Schacter 1987).

These modules will be domain-specific, that is, each one will operate only on a restricted appropriate set of
inputs, for example information about physical causation, time, space, or social relationships (see Gallistel 1998; Shettleworth 2000). A second key prediction of the adaptationist viewpoint is no organism is the proverbial tabula rasa, or blank slate. Rather, animals’ nervous systems are preorganized to process information in species-appropriate ways. Not only such specialized learning abilities as bird song (Chapter 1) but also associative learning, memory storage, attention, and problem solving as well as perception are matched to specific environmental requirements.

Thus, cognitive scientists should be seeking to understand the structure of information-processing in terms of the structure of the world. For example, Cosmides (1989) claims that the ability to solve the Wason selection task, a logical problem, reflects an ability that was selected because it helped in
detecting cheaters on social contracts. This notion predicts that people should reach the logically correct solution more often with problems about detecting cheaters than with formally identical problems about other material.

Although many data are consistent with this hypothesis, it has not gone unchallenged (Chapter 12). The
same kind of argument has been applied to experimental tests of the adaptive value of Pavlovian conditioned responding (Chapter 4). Such research is implicitly based on the argument from design: ‘‘X appears to be designed specifically to do Y; if it is, then
animals with X should be better at Y than at some superficially similar but adaptively irrelevant task, Z.’’
The evolutionary psychologists’ approach is essentially the same as the approach to cognition taken in this book.

However, it faces several problems. Some stem from
the indirectness of the relationship between cognition and fitness depicted in Figure 2.7. As Lehrman put it, ‘‘Nature selects for outcomes, not processes of development’’ (Lehrman 1970; Shettleworth 1983; Rozin and Schull 1988). Function does not uniquely determine the details of causation (Hogan 1994a; Bolhuis 2005). For instance, if the adaptive problem solved by eggshell removal is reducing predation, why didn’t gulls evolve eggshells that were cryptically colored inside?

The answer to this sort of question may lie in constraints from other aspects of the species’ biology. The way in which eggshells are produced in the gull’s oviduct may not readily allow for a change in the color of their interior, whereas gulls need motor patterns for picking things up and carrying them in foraging and nestbuilding, and these could be used equally well to carry eggshells. To take an example from cognition, many Figure 2.12. What is this? For the answer, see Figure 2.14 at the end of the chapter.

animals need to be able to return to a home to care for their young or to gain protection from predators. Thus they need a cognitive device for remembering and
relocating places, but its details may differ from species to species. For example, dead reckoning (path integration) is accomplished very differently by rats and ants (Chapter 8). Similarly, because animals are selected to care for their own offspring rather than unrelated young, species with parental care must have mechanisms for recognizing their own offspring, in some sense.

This can mean nothing more sophisticated than spending a couple of weeks stuffing food into any gaping mouth in your nest, but animals with young that run around while their parents are still feeding them
need another mechanism, such as mutual learning of identifying cues. Thus although the prediction that offspring should be favored does not tell us how a particular species recognizes its young, a closer look at the species’ biology may make functional sense of the mechanisms by which it does so. Conversely, identifying the function of a process discovered in the laboratory can raise new mechanistic questions that would not have been asked otherwise (Sherry 2005).

2.5.2 Adaptive specializations and general processes

If an ability is an adaptation to certain ecological requirements, it should vary quantitatively across species with those requirements. More spatial information to process means more capacious spatial memory (section 2.3.2); reliance on olfaction for foraging at night means relatively bigger olfactory bulbs (in birds anyway, see Healy and Guilford 1990); more complex social groups may mean better-developed
social cognition (Chapter 12). 

These statements describe adaptive specializations of
characters that species share. Such variations are readily observed in characters like beaks in birds (Figure 2.13). A bird that drinks nectar needs a long narrow beak, one that lives on hard seeds needs a beak like a nutcracker, one that tears flesh needs a hooked beak. Of course such changes are rarely confined to a single character but must be accompanied by adaptations of the digestive system, prey-catching behavior, habitat preference, and so on.

As Darwin argued, evolutionary change can be seen as
resulting from gradual modifications from some ancestral state. As a result, the characters of any given species are both unique, or adaptive specializations, and general, or shared with many other species.
Unfortunately, in the study of learning adaptive specialization has too often been set in opposition to general processes (Macphail and Bolhuis 2001). There is a historical reason for this.

Adaptive specialization was introduced into discussions
of learning by Rozin and Kalat (1971) in a landmark paper about flavor aversion learning and other newly described phenomena that seemed to reveal qualitatively new kinds of learning. For example, rats learned aversions to flavors that were followed by illness even when a single experience of illness had followed sampling of the flavor by many hours. Flavor aversion learning seemed to be comprehensible
only by thinking of animals in the laboratory qua animals rather than qua model humans or general learning machines.

In fact, conditioned flavor aversion and related findings turned out to have the same properties as other examples of associative learning, but with quantitatively special—and functionally suitable—parameters (Chapter 4). Thus they illustrate in a very compelling way how general processes of learning are expressed in a species- and situation-specific way, that is, with quantitative specializations. Just as with the debates about concerted vs. mosaic evolution of the brain, or general intelligence vs. modularity, the truth about general processes vs. adaptive specializations is ‘‘both.’’

In any case, opposing generality and specialization is biologically incorrect. Commonality and diversity are two sides of the same coin (Rozin and Schull 1988),
and one should not be emphasized at the expense of the other. People interested in general processes have tended to compare species widely separated on the evolutionary tree, for example pigeons and rats as in Box 1.3, whereas the study of adaptive specializations is associated with comparison of close relatives chosen for having different behaviors in the wild.

As Papini (2002) has argued, both approaches have
much to reveal about the phylogenetic distribution and evolution of learning mechanisms, just as they are doing with genetics and neurobiology. Thinking in terms of function and evolution, of convergences and divergences of both close and distant relatives, is a tremendously powerful tool in comparative psychology. For example, we learn in Chapter 10 that monkeys but not pigeons solve a test of transitive inference in a way that suggests they form a representation of an ordered set of items. 

That is, when exposed to training designed to teach them, in effect, ‘‘green is better than red,’’ ‘‘red is better than blue,’’ ‘‘blue is better than yellow,’’ ‘‘yellow is better than purple,’’ monkeys behave appropriately (i.e., choose red) when presented with the novel red and yellow pair and pass further tests that pigeons fail. Is this simply a mammal-bird difference, a difference in general intelligence perhaps? But asking what transitive inference might be good for in the real world suggests that it is useful for animals that form social hierarchies, regardless of whether they are mammals or birds.

And here the general study of animal behavior becomes integrated with investigations of the generality of this cognitive process in suggesting species to study. The corvids include species with and without dominance
hierarchies, thus providing subjects for one test of whether the ability to ‘‘do’’ transitive inference is confined to primates or is convergently evolved in species living in groups that need a certain kind of social intelligence (Kamil 2004; Paz-y-Mino C. et al. 2004). Thus integrating investigations of mechanistic and functional questions about cognition does not mean confusing the answers to different sorts of questions but rather developing a science in which information about how cognition may be used informs investigations of how it works.

Question 10

2. Evolution, Behavior, and Cognition: A Primer

2. 6. Summarizing and looking ahead

Answer 10

Just as Chapter 1 introduces the study of comparative cognition, this chapter introduces the study of evolution and adaptation. A claim that any character is adaptive can be tested in three ways: by modeling, to discover how well the character serves a hypothesized function; with the comparative method, to test whether variations in the character across many species are related to variations in ecology; and by experiment. Ideally two or more of these methods can be used together.

Using the comparative method requires good inferences about the phylogeny of the species being compared. Evolutionary psychologists claim that understanding how cognitive mechanisms evolved and what they are for can help us to understand how they work. However, testing evolutionary hypotheses about
cognition can be difficult because cognitive processes affect fitness indirectly, through the medium of behavior.

We have encountered three sets of contrasts in this chapter that seem intuitively to have much in common: Mosaic vs. concerted evolution, modularity vs. connectedness, adaptive specialization vs. general process. All seem to express a tension between a focus on parts with their specific properties and a focus on a whole with what its parts have in common. In the long (or maybe not so long) run, the kinds of processes they refer to may be linked mechanistically; developmental modularity is already being linked with evolution (West-Eberhard 2003; Schlosser and Wagner 2004).

In any case, the conclusion to be drawn from discussion of each of these contrasts is that the truth is usually a mixture of both. It may be human nature to
focus on only particularities or only wholes, but ‘‘It would be difficult to overemphasize the importance of agility in being able to appreciate both the modularity and the connectedness of biological organization’’ (West-Eberhard 2003, 83).

Question 11

What can we learn from other animals cognitive skills?

how do we measure cognition?

Answer 11

psychological approach - anthropocentric/lab oriented
psychologists tend to ask can animals do what humans can do?

biological approach - behavior in the wild
biologists tend to ask why do animals do what what they do in the wild?
———————————————————————————
comparing human and animal cognition:
dichotomous framework
- low-level mechanisms (attributed to animals)
automatic responses
“consciousless” responses
procedural responses
associative learning

• high-level mechanisms (humans)
  intentional responses
  conscious responses
  declarative responses

What is associative learning?
- learning about relationships between events
two types of associative learning

•pavlovian (or classical) conditioning is where associations are learned between stimuli.
for example before conditioning a dog, we know that unconditioned stimulus (food) is going to create an unconditioned response, which is salivation. we also know that neutral stimulus such as a fork is now going to trigger a salivation response (unconditioned response to neutral stimulus). during the conditioned stimulus is that we present together the fork and food which can create a conditioned response which will trigger salivation. now the conditioned stimulus (the fork) is going to trigger a conditioned response (the salivation.
•instrumental (or operant) conditioning is where associations are learned between actions and stimuli.
for example, in the Skinner Box experiment, a rat learned to associate voluntary actions pressing a lever with the consequences that follow from performing it which is obtaining a reward. the behaviour is going to become a strength through positive reinforcement (if the rat keeps pressing the lever to obtain rewards, the association is going to be stronger).

but is associative learning a simple explanation?

what we know about Pavlovian experiment:
an association between the CS and CR 
- a procedural representation
- an association between the CS and US
- a declarative representation

how can we tell the difference?

a re-evaluation of pavlovian (or classical) conditioning will involve associating the food with poison for instance, which is going to create sickness.
and now we test the response of the dog with the conditioned stimulus to the fork which has not been associated with the vomiting. so now there are two possibilities:

•salivation suggests that the dog has learnt an association between the CS and CR
- a procedural representation
•nausea/vomiting suggests that the dog has learnt an association between the CS and US
- a declarative representation

therefore associative learning might not be that simple.
therefore animals might be having mental representations of previous associations
——————————————————————————-
how do we study animal cognition and comparative psychology?

the ecological approach
this approach involves focusing on the adaptive problems of species and how it goes about solving adaptive problems.

cognitive adaptation involves:
•behavioral adaptation - perceptual and behavioural processes are organised flexibly with an individual organism making decisions among possible courses of action based on an assessment of the current situation in relation to its current goal.
•mental representation - goes beyond perception.
•flexible - allow the individual to flexibly organise perceptual and behavioural processes but also is going to allow the individual to use information flexibly in different context.

For example,
there are some species of wasps that bring dead prey to the entrance of a tunnel, then they enter and inspect the tunnel, return to the entrance and bring the prey inside. intelligent and planned behaviour - the wasp seems to be assessing the tunnel for its sustainability before they bring in the prey. but if a human moves the prey in a different position, while the wasp is inside inspecting the tunnel, when the wasp returns she will reposition the prey in the particular way she did before and reinspect the tunnel. they will continue doing so several times. this is an inflexible behavior and that probably could be tempted to argue that it has been programmed into the organism nervous system to face recurrent situations. so this would not be an example of cognitive adaptation but it is a specific adaptation.

Another example,
some species of animals have to learn about particular foods available in their environment. how to find them, acquire them and process them. have to learning about the different individuals and their groups will have to cooperate and compete with them effectively. because these skills are learned, the strategies might be unlearned or modified through experience, meeting with a change in circumstances such as finding food after flawed. this ability to flexibly adapt to changing environments or changing circumstances is a critical aspect of the cognitive adaptations.

In summary, being able to use a tool to interact with other individuals, being able to find food is an example of learned behaviours and these learned behaviours can be unlearned or modified through experience if the circumstances change.

under this approach, how do we measure cognition?
•behavioral adaptations by which an organism copes flexibly with a situation by choosing courses of action based on its understanding of the current situation in relation to its current goal.
•individual/species are understood in their ecological context
e.g., which problems they have to solve and how

under ecological problems, animals face two main types of problems, physical or technical which usually involve the presence of cognitive skills that allow animals to deal with and understand the physical environment. aids cognitive skills for dealing with the physical world (in the cost context).

Physical cognition:
apes for example, and any other animal who have to deal with finding and obtaining food, and there’s no doubt that certain cognitive skills have evolved in this context like for instance,
•spatial concept - where is food? how can I get there?
•distinguishing characteristics - which berries are eatable, and which one are baneful?
•distinguishing quantities - what tree has more fruits?
•tool use - how can I open a nut?

another problem is,
social cognition:
•communication - interpret individuals (using signals)
•social learning - predict behaviours, consequences
•perspective taking - capacity to see what others can’t
•cooperation - obtaining food, mating other individuals

physical cognition: tool use - 
peanut experiment (Hanus, Mendes, & Call, 2010)
this was carried out with great apes and also children, and the question was that they were presented with a tube and there was a peanut at the bottom. they couldn't reach with their hands and couldn't move/remove the tube to get the peanut out. so this allusion was to fill the tube with water because the peanut can float, therefore could obtain the peanut.

social cognition: theory of mind -
- knowing what others can see - deception
the experiment has two boxes, one is transparent while the other is not. if the chimpanzees understand that the experimenter can see through the transparent box, they will come and steal the food from the non-transparent box instead
- gaze following
another important problem that animals have to face is understanding what other animals can see or cannot see. this is an example of how they follow the gaze of a human. so the human is engaged looking at something, and the chimpanzee might understand what she is looking at is something interesting. so she is going to follow the experimenter’s gaze.