The Skeptics Society & Skeptic magazine

When navigating the modern world with its varied conveniences and modes of leisure, it seems that we humans are completely detached from the harsh environments that our species evolved out of thousands of years ago. Under stress, or in moments of crisis, however, the tools that our minds have evolved to deal with danger or imminent threat become quite apparent. During times like the recent global COVID-19 pandemic, when resources become unpredictably unavailable, we can turn to rather selfishly acquiring large quantities of particular products. From toilet paper rolls to baking flour, perceived essentials are coveted and cached away, hidden from other individuals, reserved for personal use in the future.

During such periods of uncertainty and upheaval, we aim also to construct meaning and a story line from the world rapidly changing around us—one by-product of which is the development of conspiracy theories. While such actions may be frowned upon in today’s society, and can be explained by hardwired behavioral reactions, they also point out the sophisticated cognitive tools that were likely critical to our evolutionary survival, indeed success, namely: recall of specific past events, future planning, the attribution of mental states to other individuals (theory of mind), a strong belief in some source of causation, and an underlying curiosity about the world we live in.

Thankfully, perhaps, we are not the only species with a tendency to cache goods when resources become scarce or when environments are risky—this is a trait we share with over 200 other vertebrates.¹ Food-caching behavior is particularly impressive among birds such as the Clark’s nutcracker. This species lives in harsh seasonal environments and can cache tens of thousands of pine seeds within a season. Remarkably, they are able to remember and retrieve the seeds with great accuracy over nine months after storing them.² The scrub jay on the other hand, caches a smaller number of more varied items, some of which perish relatively quickly (insects and olives, for example), and must therefore also keep track of the decay rates of different food items, and the passage of time, in order to successfully retrieve edible snacks.³ Are these remarkable behavioral feats potentially underpinned by sophisticated cognitive tools like our own, or can they be explained in terms of simpler, hard-wired behavioral predispositions?

Ethologists and comparative psychologists who study some of the cleverest organisms on the planet have grappled with such questions concerning the nature and origin of intelligence for decades, across a wide variety of different contexts and animal taxa. The comparative study of animal cognition has raised a number of critical questions over the years, including: Are other animals conscious?⁴ Can they “mentally travel in time” by storing specific memories and imagining the future?⁵ Are non-human animals able to attribute mental states to other individuals,⁶ and does curiosity motivate their interaction and exploration of these abstract phenomena?⁷ Ultimately, what is it about human cognition that sets us apart from other animals, and why? Trying to answer these types of questions is more important than ever. Not only does it give insight into the nature and origins of our own thinking and behavior, tackling these questions can also help us better understand, build, and predict artificial forms of intelligence, which are becoming increasingly embedded in the fabric of society and our daily lives.⁸

Though comparative cognition is a vast field, researchers are unified by a central challenge: unlocking the secrets of animal minds, which are like black boxes whose contents are neither directly visible nor accessible. Unlike work in human psychology that can partly rely on participants to report their own subjective experiences, research in animal cognition must employ creative behavioral tasks and interventionist approaches in order to test causal hypotheses about mechanisms that underlie behavior. This is the only way to tease apart hardwired responses or simpler forms of associative learning from more complex forms of cognition that could potentially explain behavior in question.⁹

Take, for example, the remarkable (and often frustrating) ability of ant colonies to identify and efficiently transport food from sparsely scattered patches in the environment to their nests. Research employing mazes has shown that Argentine ants are capable of solving fiendishly difficult transport optimization problems, flexibly finding the shortest path to food sources, even when known routes become blocked off.¹⁰ When watching individuals zealously journey out of the nest and back again, in close coordination with one another, it would be reasonable to assume that each ant had an understanding of the transport problem being solved, or that a central organizing force was shaping the behavior of the colony. Yet this feat is an example of self-organizing collective intelligence; a phenomenon that does not require a global controller, or even that the individuals be aware of the nature of the challenge that they are solving together. By adhering to simple, fixed rules of pheromone following and production, individual ants by means of only local interactions can produce complex collective behavior that does not rely upon any sophisticated cognition at all. This example highlights the need to employ carefully crafted experiments to elucidate correctly the true nature of behavioral processes.

Understanding the nature of intelligence is a tricky business but comparative psychology provides us with experimental tools that offer a window into the mind’s eye of other animals.

Initially, comparative studies of complex cognition focused primarily on other primate species.¹¹ Their close evolutionary relation to humans means they provide something of a window into the ancestral origins of our sophisticated cognition, and by comparison, the novel idiosyncrasies that characterize human intelligence (although they too have evolved both their bodies and their behavior in response to the selective pressures they have encountered since their split from us and our common ancestor). Nonetheless, it is anthropocentric to assume that complex cognition is exclusive to primates. Indeed, research on primate cognition has generated two influential hypotheses for the evolution of advanced intelligence that are applicable to a wide range of taxa. The Ecological Intelligence Hypothesis suggests that challenges associated with efficiently finding and processing food promote sophisticated cognition,¹² while the Social Intelligence Hypothesis argues that activities involved in group living, including the need to cooperate with and potentially deceive others, drive the evolution of sophisticated cognition.¹³

Over the last three decades increasing evidence has accumulated to show that a similar combination of selective pressures has driven the evolution of comparably complex cognition in other animal groups, notably the corvids.¹⁴ This group of birds, which includes crows, jays, ravens, and jackdaws, is capable of remarkable behavioral feats. These include the manufacture and use of tools for specific tasks,^{15, 16} and even the ability to “count out loud” by producing precise numbers of vocalizations in response to numerical values.¹⁷ The discovery of such behaviors points to complex underlying cognition, and given that primates and corvids diverged some 300 million years ago, it also suggests that advanced intelligence evolved independently at least twice within animals as the result of convergent evolutionary pressures.

In order to closely elucidate the nature of intelligence in animals, it is instructive to first identify natural behavior that may reflect complex cognitive processes, especially ones that can also be studied in controlled laboratory conditions. The food caching behavior of birds has proven to be a powerful model through which to investigate the nature of animal intelligence across a range of domains, including recall of past events, future planning, and the ability to attribute mental states to other individuals (“Machiavellian intelligence”). In particular, laboratory studies on scrub jays have leveraged that species’ propensity to cache a variety of perishable foods, but not eat items that have degraded. How do individual birds efficiently recover the hundreds of spatially distinct caches they make daily, given that different food items decay at different rates?

In a notable study published in Nature,¹⁸ researchers hypothesized that jays use a flexible form of memory that previously had been thought exclusive to humans—episodic memory. Episodic memory allows us to recall specific events that have occurred in our mind’s eye, and we experience these memories as our own, with a sense that they represent events that have occurred in the past. In the absence of a method to ascertain whether jays subjectively experience memories as we do, the researchers proposed behavioral criteria that would indicate “episodic-like” memory: an ability to retrieve information about “where” a unique event or “episode” took place, “what” occurred during the event, and “when” it happened. To test this, they conducted a series of experiments in which jays were presented with perishable worms that could be cached in trays at one site and non-perishable nuts that could be cached at another. The results of the experiments showed that when given the option to recover caches after a short time, the birds preferred to search for the more desirable, tasty worms, but switched to searching for the less attractive nuts after longer delays, when the worms had decayed. These experiments demonstrated for the first time that a non-human animal can recall the “what-where-when” of specific events in the past using abilities akin to episodic memory in humans.

While birds might rely on recall of specific events to successfully retrieve cached items, the initial act of caching itself is prospective, functioning to provide resources for the future when they might otherwise be scarce. This raises the possibility that non-human animals are capable of future planning, mentally traveling forwards in time to anticipate future needs that differ from present ones. However, caching may also simply be a hardwired behavioral urge, rather than a flexible response that is reliant on learning. To explore this, researchers tested scrub jays using a “planning for breakfast” paradigm.¹⁹ Over a period of six days the jays were exposed daily to either a “hungry room” where breakfast was never provided, or a “breakfast room” where food was available in the morning. Otherwise, the jays were provided with powdered (uncacheable) food in a middle room that linked the other two. Then, the birds were offered nuts in the middle room, and the opportunity to cache them in either the hungry or breakfast room. The results showed that the birds spontaneously strongly preferred to cache the nuts in the hungry room, indicating for the first time that a non-human animal can plan for the future, guiding its behavior based on anticipated future needs independent of their present motivational state.

The examples above demonstrate the ability of birds to “mentally travel in time” and form representations of their own past and future. To recover their caches successfully, however, each individual bird must also pay attention to the other birds who might attempt to steal their caches. To lessen the risk of that happening, individual birds employ a range of strategies to protect their stored food, including caching food behind barriers, out of the sight of other birds, and producing decoy caches that do not contain any edible items. To explore the cognitive processes involved in cache protection behavior researchers allowed scrub jays to cache food when alone, or while being watched by another bird. The caching birds were then provided the opportunity to recover their caches while in private, giving them a chance to re-cache the hidden food items that might be vulnerable to pilfering. Interestingly, not all birds re-cached the items most at risk of being stolen (those cached in front of the conspecific). Only those scrub jays who were experienced pilferers themselves decided to re-cache items that had been watched by another individual.²⁰ The implication is that birds who have been thieves in the past project their experience of stealing onto others, thereby anticipating future stealing of their own caches. In other words, it takes a thief to know one! This experiment therefore raises the possibility that the jays simulate the perspectives of other individuals, suggesting that like humans, they may be able to attribute mental states to others, and therefore have a knowledge of other minds as well as other times.

The approach employed in these studies highlights the utility of exploring behavioral criteria indicative of complex cognitive processes by using a carefully controlled experimental procedure. One advantage of this approach is that it is widely applicable, since it relies on externally observable behavior, rather than obscure internal states, and can therefore be used to investigate a diverse range of intelligences. Recently, comparative psychologists have started to apply these techniques to systematically investigate the intelligence of soft-bodied cephalopods—the invertebrate group comprised of octopus, cuttlefish, and squid.²¹ These remarkable animals have captured the imagination of naturalists for hundreds of years and reports suggest they are capable of highly flexible and sophisticated behaviors. For example, veined octopuses transport coconut shells in which they hide themselves when faced with a threatening predator, raising the possibility that they may be able to plan for the future. Further, the male giant Australian cuttlefish avoids fights with other males by deceptively changing their appearance to resemble that of females—perhaps they are capable of attributing mental states to other members of their species.

A coconut octopus (Amphioctopus marginatus), hides from threatening predators between a coconut shell and a clam shell. Using its tentacles, it carries the shells, while pulling itself along. Sensing a threat, the octopus clamps itself shut between the shells. (Photo by Nick Hobgood, Wikimedia)

Recently, laboratory experiments with the common cuttlefish have shown that like some birds, apes, and rodents, they are able to recollect “what-where-when” information about past events through episodic-like memory.²² Unlike other species however, episodic memory in cuttlefish does not decline with age, offering exciting opportunities to study resistance to age-related decline in cognition.²³ As with food caching among corvids, behavioral experiments with cuttlefish have also revealed prospective, future-oriented behavior: after learning temporal patterns of food availability, cuttlefish learn to forgo immediately available prey items in order to consume more preferred food that only becomes available later.^{24, 25} Presently, however, it is not clear whether this reflects genuine future planning, which requires individuals to act independently of current needs—and so presents an exciting avenue for future research.

Given the broad applicability of the experimental approach developed in comparative psychology, it is worth considering the utility of experimental paradigms to investigate the behavior of non-organic forms of intelligence. Artificial Neural Networks (ANNs) are becoming increasingly embedded in the way that we work, solve problems, and learn, perhaps best exemplified by the advent of Large Language Models (LLMs), such as ChatGPT, now ubiquitous by their use in content creation and even serving as a source of knowledge.²⁶ It is more important than ever that we develop an understanding of the behavior of these forms of intelligence. Fortunately, decades of research aimed at understanding the minds of animals has provided us with the conceptual tools needed to elucidate the processes underlying artificial behavior, and the means to build a form of artificial intelligence that is more flexible and less biased. Though reports of ANNs besting humans in traditionally complex, strategic games such as poker abound,²⁷ some have argued that these wins are often restricted to very specific domains, and that ANNs are far from displaying the general intelligence of animals, let alone humans.²⁸

This article appeared in Skeptic magazine 29.3
Buy print edition
Buy digital edition
Subscribe to print edition
Subscribe to digital edition
Download our app

Interdisciplinary efforts, however, are helping to close this gap. Inspired by research in cognitive psychology, computer scientists have incorporated an analogue of episodic memory into the architecture of ANNs. Endowed with the ability to compare present environmental variables with those encountered during specific points in the past, ANNs are able to behave much more flexibly.²⁹ Recently, influenced by classic tasks in comparative psychology, psychologists and computer scientists have collaborated to produce a competition testing the relative cognitive abilities of ANNs.³⁰ Dubbed the “Animal-AI Olympics,”³¹ this competition should help to promote the development of artificial forms of intelligence capable of mirroring the general intelligence displayed by animals, and perhaps one day, humans.

Understanding the nature of intelligence is a tricky business, but comparative psychology provides us with experimental tools that offer a window into the mind’s eye of other animals. In the future, these approaches may prove invaluable in providing insights into the behavior of artificial forms of intelligence, and one day, perhaps, into the behavior of organic life that looks very different from that on Earth.