Anthropofabrication and the redressing of memory: an embodied approach to comparative cognition

Abstract

On what basis do researchers posit that humans and other animals share cognitive capacities? We argue that such claims are not based on inherent, pre-existing similarities, but rather emerge through a two-step process, which we will call ‘anthropofabrication’. In the initial stage, embodied action-based strategies and environmental context in human studies are ignored owing to the need for measurement and quantification. Consequently, cognitive terms become disconnected from the context to which we apply them, and human classificatory cognitive terms are transformed into broad explanatory terms, assumed to be ‘species-neutral’. The second phase entails translating and applying these generalized explanatory terms to specific nonverbal animals in ways that serve to further cloak differences between animals and other species. Here, again, researchers selectively discard contextual information to facilitate the comparison with humans. To limit anthropofabrication, we should (re)acknowledge that cognitive abilities are not species-neutral and cannot be detached from embodied action, perception and their context of occurrence. We illustrate our points about anthropofabrication using the example of memory research.

This article is part of the theme issue ‘Minds in movement: embodied cognition in the age of artificial intelligence’.

And what the computer people don’t realize, or they don’t care, is we’re dancing animals. You know, we love to move around. And it’s like we’re not supposed to dance at all anymore.

~Kurt Vonnegut, interviewed by David Brancaccio on the PBS NOW show, October 7, 2005

1. One size does not fit all

Darwin’s assertion that ‘[t]he difference in mind between man and the higher animals, great as it is, certainly is one of degree and not of kind’ [1, p. 105] continues to loom large in comparative psychology. This statement encourages the search for cognitive abilities that humans share with other animals. Differences, if discussed at all, typically appear as a lack of, or at best a precursor to, ‘full-blown’ human abilities, with little recognition that Darwin also was at pains to emphasize that evolution is a diversity-generating process. Of course, such an anthropocentric goal is not wrong per se. As MacLean et al. [2, p. 224] note, ‘if we understand how cognition evolves in nonhumans, this knowledge may in turn inform our understanding of how our own species’ cognitive abilities have evolved’. But what if our dominant idea of similarity is built on shaky grounds?

Many comparative psychologists endorse a universal language of cognitive mechanisms and processes, despite considerable variation in the morphology and ecological niches of animals. They regard cognitive processes as brain-based, species-neutral and universally applicable across organisms. In order to make comparisons, researchers translate concepts from human psychology into concepts that apply to non-human animals. They deem this process of translation as a straightforward and harmless practice of generating effective experimental designs, but we think that ‘[t]he decisive moment in the conjuring trick has been made, and it was the very one that we thought quite innocent’ [3, p. 308]. That ‘decisive moment’ is the assumption that cognitive abilities are species-neutral.

As Sandis [4] has shown in his analysis of Wittgenstein’s famous statement about a talking lion, ¹ transposing concepts from human to non-human contexts is different from transposing concepts between two human cultures. Understanding other humans, whose language we do not understand and whose customs differ ‘does not require us to learn a new ability; we may currently lack their ways and concepts but already possess the ability to master them’ [4, p. 149]. For instance, a Danish person visiting Japan might not speak the language and be puzzled by the customs, but nevertheless she will share, at the very least, the same biological structure and sensory and bodily abilities. But to make sense of a lion, elephant, baboon or raven, we do need to acquire a whole new ability: we have to immerse ourselves in their ‘form of life’, to use Wittgenstein’s [3] term. Other animals have unique sensory and bodily abilities, and all live in very different niches compared with humans.

To immerse oneself in another form of life requires us to recognize and embrace complexity. By contrast, in our reading of the psychological literature, it appears that the process of translation requires the pruning of human concepts—of operationalizing terms in ways that facilitate comparison. Specifically, in order to apply human psychological concepts to non-humans animals, we have to strip them down, with certain aspects—those related to perception, action, environmental context and experience—adapted, bracketed or left out altogether. This seems justified because those aspects are seen as ‘non-cognitive’, and hence incidental to the concept of interest. This, in turn, reflects the prior assumption that cognitive processes are confined to the brain, and to the brain alone. As Rowlands [5] has noted, the assumption that mind and cognition are wholly internal is a hangover from a Cartesian view of the mind. More precisely, although Descartes’ notion of the mind as a ghostly, non-material substance has been roundly rejected by many philosophers and by virtually all psychologists and neuroscientists, the assumption of internalism (that the mind, whatever it is, is to be found within the organism) has been subject to no criticism at all and remains intact, despite the efforts of those working in the area of 4E cognition [6]. Comparative psychologists share this deep-seated assumption that cognitive processes are wholly internal and brain-based. As a consequence, they also gather limited data about what their animals do, as well as where, when and how, because these form no part of the ‘lean’ concept generated through translation. Stripping away the situational and embodied aspects of cognition thus makes it seem as though we are stripping away superficial layers of difference to uncover deeply rooted similarities. We would argue that we are, in fact, actively manufacturing superficially similar cognitive concepts, by replacing embodied, situated elements—which we see as constitutive of ‘cognition’—with an abstract, intellectualized view of these processes, so that they can be made to apply generally. We will use the term ‘anthropofabrication’ to refer to this two-pronged process of making animals appear ‘human-like’ by fabricating similarities and cloaking differences.

Anthropofabrication is related to Buckner’s [7] notion of anthropofabulation. Researchers engage in anthropofabulation when they define particular psychological capacities by reference to an exaggerated sense of human performance and only count performance at this level as ‘genuine’ possession of the skill by non-human animals. In other words, we are often wrong about our own cognitive prowess, and we judge animals according to these inflated standards in ways that serve to compound the problem. In our view, anthropofabrication precedes anthropofabulation, because debates about the definition and criteria for possession of certain abilities depend on the initial assumption that cognitive processes can be regarded as wholly species-neutral.

To mitigate anthropofabrication we have to redress—that is, re-situate and re-embody—cognitive abilities. In what follows, we first illustrate our points about anthropofabrication by discussing the history of the concept of memory and how it has been studied in humans (§2) and, as a consequence, in other animals (§3). This serves to illuminate how ‘artificial’ our current scientific notion of biological memory really is, in the sense of being both produced and constituted by human artifice and artefacts. We then move on to a discussion of how we can limit anthropofabrication by embracing within-species and between-species variation in morphology, sensory modalities and sociomaterial processes (§4).

2. Step one: fashioning memory

If you asked a psychologist, comparative or otherwise, ‘what is memory?’ they would probably bring up something about storage and retrieval—much like putting your winter jacket in a box during summer and fetching it when winter returns. For instance, Shettleworth [8, p. 210] broadly defines memory as a cognitive ability that ‘deals with how information is stored, retained and retrieved’. Given developments in memory research over the past three decades or so, researchers would concede that your jacket may not be the colour you anticipated, or it may look different from how you imagined it. That is, our memories are known to be susceptible to distortion, alteration and inaccuracy [9–11]. They would also say that memory comes in many shapes and forms, such as episodic, semantic, procedural and autobiographical memory. The increasing differentiation of concepts also draws attention away from situated activities but here we will not detail how they do so. Instead, we will stick with the fact that the storehouse formulation of memory diminishes attention to environmental structures and embodiment. Thus, memory is strongly regarded as a means by which humans and other animals store a record of past events, albeit an imperfect and revisable one.

As Danzinger [12] has argued, we tend to presume that our current scientific notion of memory—of (imperfect) storage and retrieval—points to what memory ‘naturally’ is, and that it evolved through natural selection because it provided an adaptive advantage. However, consider an example that challenges this presumption: in Ancient Greek cultures, stories and poems were transmitted orally. Recitation was a performative activity enriched with rhythm, intonation, gestures and bodily movements. These elements played a crucial role in aiding memorization. Through these expressive means, the storyteller reconstructed the narrative with each telling. Today, we recast these embodied techniques in terms of scaffolding for memory, as if the bodily techniques serve only to elicit the recall of memory in the head. For example, there is work showing that recollection is enhanced when the movements that participants make during recall align with their movements during the initial learning process [13]. This is a slow reintroduction of the body into an activity that is and always has been fully embodied and situated.

Such a view is at odds with the view of memory at the time. The storyteller’s gestures and actions were not considered mere by-products of memories; rather, they actively constituted the process of remembering. The strict separation between body and mind, and between past and present, did not hold up: ‘[N]o distinction between reproduction and composition can be detected in the earliest Greek literature that refers to Mnemosyne [meaning: ‘remembrance’], the exercise of memory as an activity. […] Early on, remembering means listening to a voice’ [12, p. 29]. The isolation of memory as a distinct (cognitive) process—as a thing that we possess, rather than an activity in which we engage—occurred much later with the invention of writing.

By writing down words, the act of speaking and its ‘content’ are separated, and written words therefore take on a quasi-permanent existence, subject to inspection by numerous individuals [12, p. 33]. Consequently, as humans embraced writing, they gained the ability to compare recollections of people with recorded events, which allowed them to assess how accurate these recollections were [12, p. 50]. Interestingly, writing became a metaphor for memory itself, leading people to believe that memory resembled the storage of words [12, p. 50]. With the advent of computer technology, computers and the language of information-processing also became memory metaphors. However, these metaphors slowly lost their metaphorical ring and became accepted as ordinary descriptive terms [12, p. 44]. Examples of such ‘dead metaphors’ are information, input, output, encoding, decoding and content. This information-processing language has led to ‘a closed conceptual world in which anything that was not expressible in the language of information processing remained unseen, unrecognized and unexplored’ [12, p. 53]. These terms, and the experiments they give rise to, thus warranted the omission of the contextual, performative elements that for the Ancient Greeks were very much part of remembering.

The prevailing image of memory as the storage and retrieval of ‘information’ was further substantiated by Hermann Ebbinghaus, the renowned psychologist, who brought memory into the lab in an attempt to quantify it. He asked research subjects (often just himself) to learn nonsense syllables and word lists and then to recall them later. Because he could compare these syllables and words in his lists, he could establish whether the participants’ memories were accurate and how much information was retained over a given time frame. However, ‘[w]ithout the use of fixed, recorded materials neither the measurement of memory performance nor the standardization of experimental conditions would be possible’ [12, p. 174]. The use of nonsense syllables, moreover, suggested that the ‘contents’ of memories were isolated ‘snapshots’ of the environment. This content was thought of as passive ‘input’ to the senses, and the performative side of remembering was left out ‘by simply not collecting such information’ [12, p. 131] in order to get to ‘memory proper, defined as “mere retention”’ [12, p. 128].

Ebbinghaus’ participants were not really taking in information passively. Their urge for action, even if minimal, and for discovering meaning within apparent nonsense, could not be suppressed. Drawing from their prior experiences, they added rhythm or sought meaningful patterns anyway. As Danziger [12, p. 135] puts it, ‘Ebbinghaus and those who followed in his footsteps had chosen material with poor intrinsic organization that could be remembered only by intentionally imposing some organization of one’s own’. While later research replaced nonsense syllables with words and word pairs, the foundational premise of memory research has remained intact [12]. At the time, Ebbinghaus’s findings were interpreted to reveal the cognitive mechanisms that underpin every instance of what we call ‘remembering’.

What Danzinger’s [12] detailed historical perspective highlights, then, is that we have come to think of remembering as storage and recollection in our heads because we have been studying and talking about remembering on this basis—and so increasingly confining it within our skulls along the way. Danziger [12, p. 5] has rightfully questioned whether twenty-first-century laboratories are investigating a universally shared and generic ‘human memory’ or rather a socially influenced manner of functioning that evolved over time owing to the gradual development of literacy. Similarly, Hutchins [14] has pointed out that much of modern cognitive science involves the study of our sociocultural tools, rather than our cognitive processes.

We can go further and bring in Alva Noë’s [15] argument that, as human beings, we are not merely integrated with our cultural products but are deeply entangled with them. Using the examples of dancing, seeing and speaking, Noë [15] argues that the ways in which we have been organized and reorganized by our cultural practices—choreography, pictorial art, writing—have been going on for so long that any attempt to investigate our ‘natural’ impulses to dance, see and speak, that is, our unadulterated ‘human nature’, would require the impossible: we would need to ‘go back to Eden, that is, go back to a make-believe prehistory’ [15, p. 31]. We see memory as an entangled concept in this sense, too. The material and social practices that we have used to support and engage with memory—literacy, external storage devices, our scientific methods for studying memory, depictions of memory in oral story-telling, mnemonic devices such as ‘memory palaces’, written fiction and other forms of representational and visual art—have now become so fused together with whatever biological capacities we possess, that this ‘entangled creature’, to use Noë’s [15] words, now just is the human form of memory that we study scientifically: we are creatures capable of storing and producing accurate recollections of the past, and for whom a more abstract, disembodied form of memory is part and parcel of human life.

In short, this form of human memory—the storage and recollection of past events—is an entangled skill, subject to further entanglement. It is not a fixed and general biological function. However, the view that psychological studies of memory are geared towards probing such a universal and general biological function is widely accepted and has seeped into other fields, including comparative psychology. Generic human memory has become generic animal memory; as such, it has been transformed from a typically human attribute into a species-neutral attribute. To illustrate this latter point, we will discuss a specific type of memory that researchers have explored extensively in Western scrub jays, among other animals: episodic memory.

3. Step two: sustaining a sense of similarity

In 1972, Endel Tulving wrote that ‘[o]ne of the unmistakable characteristics of an immature science is the looseness of definition and use of its major concepts.’ [16, p. 381]. Dissatisfied, he took steps to resolve this and made the now-famous distinction between episodic memory (memory for personal experiences) and semantic memory (memory for generalized facts). It is also perhaps worth noting that, if we again take a historical perspective, we can see how the term ‘semantic memory’ also reflects the ideas of Ebbinghaus, where the world is thought of as a vast array of ‘isolated facts’ that animals can learn to retain and is devoid of experiential aspects—an image that is very much shaped by experimental practices and storage technology (see [12, p. 174]). Tulving [17] attached the term ‘autonoetic consciousness’ to episodic remembering. This term indicates conscious recollection of a past event, an ability for ‘mental time travel’. Although by this time ‘[t]he lesson was taken that distinctions among memory systems were conceptual and classificatory, not causal and explanatory, […] in practice they were never treated as mere heuristics’ [12, p. 175]. Moreover, while conscious experience and contextual variables of the remembered event were explicitly integrated into the definition of episodic memory, the dynamic, context-dependent behaviours exhibited during remembering were still overlooked.

Taking on Tulving’s distinction, Clayton & Dickinson [18] wanted to test whether Western scrub jays (Aphelocoma californica), a North American bird known to cache various food items that vary in decay rate, also possessed episodic memory—or rather, episodic-like memory (the suffix was added to evade questions about autonoetic consciousness, which, understood as a private experience, could not be assessed empirically). Episodic-like memory was thus defined as ‘what–where–when–memory’: memory for what happened, where it happened and when it happened. Clayton and Dickinson hypothesized that episodic-like memory may have evolved in this species, not only to enable birds to retrieve caches effectively but also avoid those caches that were ‘past their sell-by date’.

The birds were placed in an experimental set-up that allowed them to cache both meal worms, a highly preferred but also highly perishable food, and less preferred but longer-lasting peanuts. Crucially, after a 7-day delay, the birds recovered the peanuts right away, despite these being less preferred as food, and avoided the preferred meal worms. As the meal worms, but not the peanuts, would decay over a 7-day period, the authors inferred that the birds remembered not only what they had cached and where, but also when they had done so, such that they recovered only the non-spoiled food. The authors concluded: ‘[T]he cache recovery pattern of scrub jays fulfils the three, “what,” “where” and “when” criteria for episodic recall and thus provides, to our knowledge, the first conclusive behavioral evidence of episodic-like memory in animals other than humans’ [18, p. 274].

The design of the study itself was elegant and clever. However, the researchers were engaging in anthropofabrication even before their experiments had started. They did not question whether episodic memory in humans could, in fact, be translated to non-human animals. Episodic memory, for them, was a generalizable, species-neutral process, which could be stripped of the features they deemed human-specific ‘add-ons’ in order to isolate its ‘core’ features. But let us take a moment to consider here the key human ‘add on’: the ‘autonoetic’ component of episodic memory, the ability to relive a past moment. As with memory in general, our notion of episodic memory may be entangled—that is, intertwined with sociomaterial practices and technologies, such as our reliance on clocks, calendars and other time-keeping methods that together with our communicative abilities, enable a sense of past events as past events, independent of our current goals or engagements. In other words, these technological artifacts and linguistic practices enable us to separate time neatly into the past, present and future and locate personally experienced events on a timeline. Does this separation also exist for animals that lack these means? If we answer ‘yes’ to this question, then we seem to be making the same mistake as claiming that the Greek storyteller was retrieving pre-existing information, rather than engaging in a reconstructive performance in which the distinction between past, present and future made little sense.

On our ‘entangled’ view, our sociomaterial environment cannot be taken out of the equation if we want to understand human episodic memory. However, on the dominant cognitivist view, situational cues or sensory–motor strategies are considered to be confounding factors (§2). Much as Ebbinghaus considered memory to be a purely ‘cognitive’ internal process, comparative psychologists similarly consider episodic memory as a purely brain-bound internal phenomenon. In order to establish episodic memory in their study animals, they therefore have to exclude alternative explanations by which animals are able to solve a memory task using occurrent cues. To this end, environmental features that animals can smell, see, feel, hear or even taste are treated as confounding variables that can be neutralized through experimental routine and standardization.

Because the jays passed the experiment, and the experiment was set up in such a way that passing indicated the possession of episodic-like memory, the researchers claimed a similarity between humans and scrub jays. In contrast, we suggest that, by first ignoring the impact of the human environment of episodic memory (thus seeing them as species-neutral, brain-based capacities) and then accounting for ‘confounds’ in the animal’s environment (e.g. olfactory or visual cues, solar cycles, circadian rhythms), the researchers were actually generating a sense of similarity as their experiment proceeded. In other words, while it seems as though we are uncovering a similarity, a case of evolutionary convergence, we are actually selectively ignoring or ‘cloaking’ species-specific, situational influences on the capacities of both humans and jays—all owing to cognitivist assumptions about species-neutrality. We discuss species-specific ‘solutions’ to these experimental tasks in the next section.

We have focused here on one specific and well-known case, but this is not an isolated incident: the species-neutral assumption about episodic memory still permeates research on episodic memory in non-human animals, including cuttlefish and bottlenose dolphins [19–22] We think that, as long as we treat human forms of memory as species-neutral capacities, we will continue to anthropofabricate. So, what is the solution?

4. A solution: redressing memory

Insights about and analyses of the behavior of worms need to be developed with the same methods and concepts used to understand human beings—without attempting to reduce humans to worms, or turn worms into miniature cylindrical human beings.

~Reed [23, p. 96]

To tackle anthropofabrication, we need a notion of memory that embraces, rather than masks, embodied action-based strategies and between-environment variation, and accordingly, acknowledges species-specific differences. That is, we need an approach that welcomes those aspects of memory that are currently cloaked by our species-neutral view and our dominant research practices.

Such an approach can be found in ecological psychology, where the term ‘memory’ captures the altered relationship between an animal and its environment without the need for ‘storing’ information. Put differently, remembering is a generic term for the things that organisms do at an earlier time that facilitate their future encounters at a later time. (We could also call this ‘perceptual learning’ as well as or instead of ‘remembering’, but stick with the latter term here to highlight anthropofabrication in cross-species comparisons.) As Palatinus & Michaels [24, p. 25] put it, ‘the consequence of personal experience is not that the old animal has new knowledge, but that it is a new animal that knows better’. Or as Danziger says, remembering results in ‘some “tuning”, in our disposition to see things in a particular way when encountered in a certain kind of context’ [12, p. 258]. We will discuss how this ecological notion of memory allows for better comparative memory research by highlighting two elements: the wealth of the stimulus and cognitive structuring, respectively. Note that these elements cannot really be separated: identifying the cognitively relevant environment hinges on considering the animal’s embodiment. That is, our task here is to understand more fully the opportunities that an environment is likely to offer an animal possessing a specific array of attributes. We can use these to categorize similarities and differences between species, as well as between individuals within a species.

(a). Patterns everywhere

The term ‘wealth of the stimulus’ signifies that the environment provides far more and richer information than is usually assumed. Comparative psychologists typically accept that animals only receive direct information from proximal sources. Everything beyond proximate sources, on this view, demands ‘cognitive’ processing. As previously mentioned, comparative psychologists strive to uncover ‘cognitive’ capacities that are independent of perception and context, and for that reason they mask immediate cues as part of the experimental procedure. However, the wealth of information available to animals in such experiments is rarely recognized. For instance, covering up the smells emanating from a peanut and a decaying worm does not yield a room devoid of scent. If nothing else, experimenters themselves give off an odour, and they tend to visit regularly, although the word ‘presence’ often obscures these facts: as Despret [25] observes:

[If] the scientist’s body is evoked, it is never for itself nor is it named as such: when seeking the body, we are offered a surprisingly abstract concept: the ‘presence’. This abstract term—most of the time under the guise of the ‘presence of the observer’—while referring to the body, actually conceals it. It conceals what the actual and concrete ‘presence’ is for the animals: the space the so-called observer’s body occupies, the body which moves, which walks, bears and diffuses smells, makes noise, follows, and does everything a body may do—including what we don’t know our body may do since we are so unaware about what it is capable of, but which animals may nevertheless perceive [25, p. 52]

So, for example, and similar to the case of Clever Hans, ² the experiments described above could well involve unconscious cueing, despite efforts to remove such ‘confounding’ variables. During initial phases of the experiments, the birds may become sensitive to the correlation between certain smells (or other sights and sounds) in the room—be they of the experimenters or something else—and the decay rate of worms [26]. In other words, the ability to remember when something happened could be tied to events in the experimental room. Experience would teach the birds such relations, similar to the way that dogs can learn to perceive ‘the passage of time’ by the decrease of scents in the environment [27].

Animals in experiments have plenty of time to familiarize themselves with the smells, sights and sounds in the testing room and can learn to focus on the relevant environmental contingencies. The behaviours that might indicate this process of perceptual learning—such as the animals’ movements between trials, their observations, and interactions with their surroundings—can all be systematically and rigorously collected during experiments. However, such elements are not currently collected as data because they are considered irrelevant to episodic memory, which is defined in isolation from the body and environment.

Humans also exploit relations among events. Such events may include self-induced ones, such as adding rhythm, chunking variables or entering variables into some other kind of meaningful organization based on the person’s past experience [28]. These human strategies, like the jays potentially relying on higher order olfactory patterns, are ways to connect past experience to the task at hand. When researchers attempt to exclude certain perceptual and motor ‘cues’ in order to more effectively infer the structure of internal cognitive processes, they are actually cloaking the environmental patterns that animals’ perceptual systems can exploit to solve the experimental task. Rather than figuring out how, where and when animals do what they do, they need the animals to act in a certain way, such that they accord with the way experiments in comparative psychology are carried out.

In addition, while psychologists tend to throw all caching behaviour on the same pile—given that a capacity for ‘memory’ is thought to underlie these behaviours—we suggest that caching and retrieving worms and peanuts may involve different processes of perceptual learning. Jays potentially learn to perceive relations between the edibility of worms and other opportunities for action that are available over similar timescales. Here, we are thinking of affordances as temporarily available opportunities for action: a worm is edible for a given period of time, and after that, it no longer affords eating [26,29]. The same could be true for the retrievability of peanuts, which may become associated with affordances that are available over relatively longer time-scales than affordances associated with edible worms (see figure 1). These relations can be studied systematically by varying (rather than excluding) the availability of information, in the manner suggested in [30, p. 25]: ‘Does the animal behave differently with respect to that thing under circumstances in which it is likely to have access to information about it, and does it behave in an undifferentiated manner when such information is unavailable?’ This systematic variation approach was also employed, for instance, by Darwin in his analysis of worms [23,31].

Figure 1.

Nested organization of affordances in jays’ caching activity. The top tier contains ecological patterns that occur over relatively longer timescales, while towards the bottom patterns are to be found over shorter timescales. Short-term patterns are nested within longer-term patterns. Although this figure describes ecological patterns, these patterns encompass several affordances. Roughly speaking, a peanut can be edible or inedible, and this changes over time (and perhaps we could also specify affordances across this time span, such as an optimal edibility). Daily temperature also affords various behaviours. Higher temperatures during the day, for instance, could ‘nest’ several other patterns, structuring daily activities. A lingering smell, together with a certain extent of mould growth, could approximate the time it takes for a worm to decay. A certain degree of fluctuation in humidity could roughly double the time it takes for a worm to decay, for example. Jays become experts at learning how these patterns of affordances are related to each other throughout their lifetimes. It is the goal of researchers to find out what affordances animals are exploiting to achieve their feats.

By systematically varying the affordances or processes to which the birds have access, researchers can determine the relations or higher-order patterns on which jays may rely for specific tasks. In other words, it is an effort to fill in the spaces currently occupied by question marks in figure 1. These could be affordances related to mould growth or plant growth, specific lingering smells, temperature or humidity changes, behaviours of conspecifics and so on.

Such a research paradigm would grant animals some agency to solve problems of their own accord, allowing genuine similarities between humans and other animals to manifest. For instance, we could ask if and how a jay and a human are able to recover something that they hid a week ago. We can look at the way that they potentially exploit higher-order (visual, olfactory) relations in their environment in order to achieve this. That is, we can see how animals (including humans) solve similar tasks through different means, and incorporate embodied differences and sensorimotor strategies into our data collection protocols. For humans, this may indeed include technologies such as clocks and calendars but for the sake of comparison, it may make more sense to focus, for instance, on visual and olfactory patterns in the environment. If this seems far-fetched, researchers have found that humans are actually much better at tracing scent trails than expected, when pressed to the ground. They will even automatically employ the characteristic ‘zigzagging’ patterns also found in rats, dogs, cockroaches and honeybees, among others [32,33]. In some human cultures, moreover, smell trails are very much an integral part of hunting [34,35]. This counterintuitive point illustrates that our image of ourselves should not be used to guide our research practices.

To be sure, we are not suggesting that animals (and sometimes, humans too) are employing ‘simpler’ strategies than researchers currently claim—that is, we reject the false dichotomy drawn between ‘mere’ associative learning or some other ‘non-cognitive’ interpretation and more complex ‘cognitive’ processes (which, incidentally, is itself an incoherent distinction, given that according to [8], association-based processes fall under the definition of cognitive—and both kinds of explanations ignore important variables that make up the wider organism–environment system). Rather, we advocate for a more evolutionarily appropriate, embodied, embedded notion of what cognition is (and in this case, what memory is) and how it operates in the natural world. So, instead of assuming that animals can reason in a ‘human-like’ way (full-blown or otherwise) whenever they display a capacity that we think humans possess, we should consider that we are more ‘animal-like’ than we like to think, in the sense that our own capacities incorporate our own unique bodily and environmental resources. Of course, this was exactly our point about the entangled nature of human memory in the first place.

(b). Stretching memory

Besides relying on the wealth of available information, animals also actively structure their environment in order to remember. Recall how Ebbinghaus confined his participants to experimental rooms and shrank their world to a table, chair and some nonsense syllables. The possibility of using sensorimotor and situated strategies for remembering was almost impossible (although as noted above many subjects tried their best), and it is no surprise that a particular form of memory was made manifest. Most (if not all) experimental studies take place in such confined settings, and so the ability to remember—whether by humans or non-humans—is also necessarily limited. The ‘real world’ is nothing like this and, given our perspective that remembering is tied to its present context, research findings generated in experimental settings will not, therefore, easily translate to ‘the wild’: remembering in the laboratory is a different species of remembering from that which occurs in richly stimulating environments that afford free movement. Critically, however, we could exploit experimental situation in ways that would point to the strategies likely to be employed in the wild.

Before we discuss non-human animals, let us look at some of the things that humans do ‘naturally’ to remember. We keep things in particular places as an integral part of memory—we distribute the process into the environment [36]. Putting objects in specific places turns ‘conceptual’ tasks into more effective, and easier, perceptual ones: if we wish to remind ourselves to eat more fruit, we will place the fruit bowl on the tabletop, not in a kitchen cupboard. If we always put our keys in a specific place, we can more easily remember them on our way out of the door. We charge our phone every evening, in the same place, so that we do not forget to do so, and so we know where it is when we need to take it with us. As Kirsch [36, p. 33] argues, ‘Experts find sufficient cues in the situation to trigger a known rule without halting the activity in order to consciously and analytically take stock of the situation and reason or deliberate about a solution’. These strategies are not necessarily deliberate, and ‘even though we often do not realize that we are structuring our workplace to help us keep track of processes and a host of other useful functions […] we should not assume that such cognitive or informational structuring is not taking place all the time’ [36, p. 37].

Why would animals not employ similar strategies when given the leeway to do so? Like humans in memory experiments, the scrub jays in the episodic memory experiments are confined to a particular (and impoverished) space, so their opportunities for action are limited, just as the laboratory limits human remembering. In many cases, such impoverishment can reveal animals’ resourcefulness, or indicate that particular information is not essential for the task at hand. For example, given that jays will return to the correct caching sites despite the removal of mealworms and peanuts, we know that their cache-recovering abilities are not dependent on the smell of the cached food. However, this finding does not imply that, under normal circumstances, the birds do not use the smell of cached food as a cue, nor does it rule out the possibility that they are using other environmental scents to register the state of cached food in the experiment (as we detail above).

Whichever way we slice it, the fact remains that animals’ action opportunities are restricted in an experimental context, and thus we have no means of discovering whether remembering involves action-based strategies for creating a more cognitively conducive environment. In the wild, jays are able to cache in particular places that maintain particular relations to each other, perhaps at particular times, and that are close to landmarks or other meaningful places, and so on. Such strategies are often seen as a means of ‘off-loading’ the cognitive burden onto the environment—that is, they acts as ‘scaffolds’ for internal memory. However, we suggest that such behaviours are the parts of memory that get masked in the experimental context by designating them as (non-cognitive) ‘confounds’ that obscure the operation of ‘pure’ memory.

Navigation studies with rats offer a similar example. Rats leave odour trails as a way to recall places they have visited before. Relying on scent is often contrasted with the use of memory, and researchers who are interested in the ‘minds’ of animals will see the use of odour trails as akin to ‘cheating’. Leaning on the environment in this way obviates the need for brain-based representational processes to solve the problem, and indicates that the animal is somehow solving a given task in a ‘non-cognitive’ fashion [37]. In other words, occurrent perceptual cues of any kind are viewed as potentially interfering with the research goal—that is, they obscure our ability to get at how animals really think (see also [38] and comment by [39]).

From a cognitive structuring perspective, on the other hand, scent trails are a means of rendering a wholly unnatural environment—one that appears similar in every direction—easier to navigate in the future. As humans, we are inclined to see space as an empty container, so that leaving behind an olfactory cue represents an addition to that environment. But the world of rats (their Umwelt [40]) is a scent-based world, and rats, by leaving scent cues for themselves, are simply remembering in the way that rats remember: leaving scents is the way in which they come to know, to cognize, their environments. Researchers see these scent-based strategies as non-cognitive, we think, because they are unaware of the perceptual strategies that humans also use (often unknowingly) to recall locations at later times. Our use of trails is more often visual (a well-trodden path, a sequence of post-it notes and book-marks pasted into a book) but trails they are, and they are integral to our ability to remember and allow the past to influence the present.

(c). Memory in full regalia

While necessarily brief owing to space limitations, our comments on cognitive structuring and exploiting the wealth of information (e.g. exploiting higher-order patterns) in the environment nevertheless allow us to reiterate our points about Ebbinghaus’s participants and the episodic memory studies on jays: you cannot take movement or context out of memory (at least you cannot do so without generating a distinct form of ‘scientific’ memory found only in the laboratory). Situated, sensory–motor strategies have been set apart as somehow ‘not memory’ to justify the search for ‘species-neutral’ cognitive mechanisms, on the model of what literate humans can achieve under highly specific conditions. The sense of similarity that we have fabricated by cloaking differences thus starts with the faulty image that we have of ourselves: by choosing to focus on only a narrow aspect of what humans are capable of doing within an overtly scientific context, and a priori excluding several active, perceptual strategies that, even if we try to get rid of them, we cannot but help employ. In all cases, we do something to remember something later. In this sense, there is no ‘passive’ sensory input that is stored and later recalled. All remembering is active and memory is therefore an activity—it is just that sometimes we just do not know the things that we do.

Finally, although we have couched our discussion in terms of ‘memory’ as a distinct faculty, we should make clear that our approach is much more task- or action-oriented—remembering where we left our keys is a different task from remembering how to make an omelette, or remembering the lyrics to a favourite song. The same goes for the activity of animal memory. This means that each task will require several classificatory types to describe, and it is impossible to ‘isolate’ any of these capacities in order to show that animals really possess it.

5. Conclusion

Our scientific interest in what occurs ‘in the mind’ of humans and animals has yielded a long-standing neglect of what occurs ‘outside the mind’—or rather, of the idea that mind could be more broadly conceived, and regarded as something that animals do (they engage in ‘minding’) rather than something they possess. We recognize the complexity entailed by working with non-human animals, however, and by no means wish to deny it. Rather, we want to encourage researchers to acknowledge this complexity, rather than selectively ignore it; to embrace variability, rather than trying to iron it out.

Consequently, we think that the way most experiments are currently set up is not helpful. These setups usually attempt to distinguish ‘cognitive’ from ‘association-based’ explanations. Experimental rigidity and standardized controls are undertaken to reduce ‘irrelevant’ variability and the influence of so-called confounding variables, or at least to diminish them. In doing so, we are trying to wrangle resistant animals into ill-fitting clothes. Avoiding anthropofabrication, however, does not mean simply making alterations in the clothes themselves so they fit better—taking them in here, adding extra material there. We simply cannot assume that all variability is eradicated once we put controls in place; variability will still be present, and ignoring it will not solve the problem. A larger organism–environment system is always, already in place.

Rather than excluding certain parts of the environment a priori because they are generators of ‘noise’, we argue that researchers should attempt to discover cognitively relevant environments through their experiments by systematically varying the availability of affordances; these environments—and the distinctive manner in which animals can act on them and enact strategies within them—must play an integral role in the generation and testing of novel hypotheses. For instance, researchers could actually create higher-order patterns between certain smells and the decay rate of worms, and later remove this olfactory pattern. By measuring response latency, for instance, they could then uncover whether animals are sensitive to these higher-order patterns. In addition, constructing ethograms could enable researchers to find patterns in the alternative strategies employed by jays when certain higher-order patterns suddenly become unavailable to them. To stick with our metaphor: we should let the animals themselves show us what clothes they like to wear. This approach will also allow genuine similarities to manifest in how animals exploit environmental features to solve the tasks with which they are faced. The idea of ‘trails’ that we discussed is one example. Whether by vision or scent, laying down a trail is a way to make homogenous environments unique and recognizable. Whether these trails are ‘natural kinds’, or whether they ‘really’ exist, is not important. What is important is that our terms allow us to ask new questions, to see things that we could not or did not see before. Rather than masking species-specific strategies, seeking the cognitively relevant environment through the recognition of embodied strategies offers a level playing field for comparison, and one that does not privilege humans.

The difference in mind between humans and all other animals (not just the ‘higher’ ones), ultimately, is both a difference in degree and a difference in kind. All remembering is about improving engagement with available affordances. However, each animal relies on different affordances, given the differences in their bodies, senses and niches, making every instance of remembering species-specific. Even the entangled skill of storing and recalling ‘information’, unique to humans, remains a fully embodied, situated way to improve how we encounter the world.

Ethics

This work did not require ethical approval from a human subject or animal welfare committee.

Data accessibility

This article has no additional data.

Declaration of AI use

We have not used AI-assisted technologies in creating this article.

Authors’ contributions

B.V.W.: conceptualization, writing—original draft, writing—review and editing; L.B.: conceptualization, funding acquisition, writing—original draft, writing—review and editing.

Both authors gave final approval for publication and agreed to be held accountable for the work performed therein.

Conflict of interest declaration

We declare we have no competing interests.

Funding

This work was funded by NSERC Discovery Grants and Canada Research Funds awarded to L.B.