Cognitive science today increasingly is coming under the influence of embodied, embedded, extended, and enactive perspectives, superimposed on the more traditional cybernetic, computational assumptions of classical cognitive research. Neuroscience has contributed to a greatly enhanced understanding of brain function within the constraints of the traditional cognitive science approach, but interpretations of many of its findings can be enriched by the newer alternative perspectives. Here, we note in particular how these frameworks highlight the cognitive requirements of an animal situated within its particular environment, how the coevolution of an organism's biology and ecology shape its cognitive characteristics, and how the cognitive realm extends beyond the brain of the perceiving animal. We argue that these insights of the embodied cognition paradigm reveal the central role that “place” plays in the cognitive landscape and that cognitive scientists and philosophers alike can gain from paying heed to the importance of a concept of place. We conclude with a discussion of how this concept can be applied with respect to cognitive function, species comparisons, ecologically relevant experimental designs, and how the “hard problem” of consciousness might be approached, among its other implications.
Cognitive neuroscience arose in the 1970s, based on a convergence of experimental psychology, computer science, and neuroscience, with the goal of understanding how a machine with the physical properties of the brain could produce specific behaviors when given specific inputs (Kosslyn & Andersen, 1992). Incorporating approaches from artificial intelligence, robotics, neural networks, and other relevant disciplines, it crystalized into a set of assumptions that came to represent what is often referred to as the standard, or orthodox, computational view of cognition. Those assumptions are dominated by a representational model of brain function, in which a version of the outside world is projected into the brain, cognition is computation over neural activation patterns, the processing architecture is largely modular and operates in a context-independent and substrate-neutral manner, explanatory strategies typically reference inner brain states, neural processing is separable from perception, and the cognizant entity is a detached observer (Engel, 2010).
Beginning in the late decades of the previous century, a variety of ideological and methodological approaches began to present alternatives to the standard model. Those perspectives assume that the subject of cognition is an agent immersed in the world; the processing architecture is highly dynamic, context sensitive, and captured best by holistic approaches; and cognition is a whole-body (embodied) experience immersed and inseparable from its environment, in which organisms generate (enact) meaning through their bodily action (Di Paolo, Rohde, & De Jaegher, 2010; Engel, 2010).
In this review, we will address what these alternative perspectives have contributed to the ongoing evolution of cognitive neuroscience, evidence supporting these perspectives, a resulting focus on the centrality of place in animal cognition, and the ways in which neuroscience research can be enriched through these new points of view.
ALTERNATIVE INFLUENCES IN COGNITIVE NEUROSCIENCE
Challenges to the standard model include perspectives from various frameworks that include cognitive ecology, embodied cognition, enactivism, neurophenomenology, and neurodynamics, among others. Each will be considered in turn.
Mettke-Hofmann (2014) described cognitive ecology as a discipline that integrates cognition, ecology, and neurobiology. Much of what is considered under this heading is often referred to as “ecological psychology,” but we use the former term to include a more diverse picture of how cognitive abilities have evolved in conjunction with other adaptations to environmental challenges and what the implications are for comparative studies of cognition and consciousness.
Ecological psychology traces back to the writing of Jakob von Uexküll (1926), who noted that organisms experience life in terms of species-specific, spatio-temporal, “self-in-world” subjective reference frames. Gibson (1979) introduced the field of “ecological psychology” and coined the term “affordance” to refer to properties of environmental structures that provide opportunities for action to organisms in that environment. The organism perceives that, and only that, which the environment affords (Koch, 2012; Gibson, 1977).
Inherent in the nature of cognitive ecology is the concept of extension. Silberstein and Chemero (2012) describe extended cognitive systems as heterogenous, composed of brain, body, and niche, nonlinearly coupled to one another. The phenomenological world of experience is neither in the head nor in the external world—it is fundamentally relational. As Thompson (2007) wrote, “The roots of mental life lie not simply in the brain, but ramify through the body and environment. Our mental lives involve our body and the world beyond…and therefore cannot be reduced simply to brain processes within the head.”
The tenets of cognitive ecology have some clear evolutionary implications. Animals are adapted to their environments because they have evolved under the constraints of those environments (Dennett, 2017; Rosati, 2017; Anderson, 2014; Willemet, 2013; Gibson, 1979). Adaptive behavior emerges from a complex balancing act that incorporates neural, bodily, and environmental influences (Clark, 1997a). Feinberg and Mallatt (2016) have provided strong arguments for the claim that consciousness is more widespread phylogenetically and is generated by more diverse neural substates than is generally believed and in fact has evolved independently several times.
Voices of caution have been raised against an overemphasis on the contemporary ecological approach to animal cognition. However, some of that skepticism seems to be restricted to the specific view that there are no qualitative differences in cognition between animal species in the processes of learning and memory (Macphail & Bolhuis, 2001; Macphail, 1996). It may well be that some cognitive processes underlie physiological/neuroanatomical states that have deep roots because of needs shared by all animals (such as learning and memory), whereas others are differentiated and derived because of environmentally specific adaptations.
The concept of embodied cognition has its roots in the phenomenological tradition that traces back to the early 20th century philosopher Edmund Husserl, who saw experience of the world to be constituted within the consciousness of each person (Husserl, 1936). More than anyone, the mid-20th-century French philosopher Maurice Merleau-Ponty anchored the phenomena of perception in the phenomenology of the lived body (the body as it experiences and is experienced). Perception, he argued, is the primordial fact of experience. The whole body is the locus of perception and of all experience (Merleau-Ponty, 1945). The whole body, with its entire sensory–motor capacity, is seen as the substrate for mind by many authors (Fuchs, 2018; Adams, 2010; Di Paolo et al., 2010; Hutchins, 2010).
Central to the concept of embodiment is a view of the brain, body, and world as “united in a complex dance of circular causation and extended computational activity” (Clark, 1997b). This requires abandonment of the idea that the mental realm is distinct from the realm of the body (Bruineberg & Rietveld, 2014; Engel, 2010). For Humphrey (1992), the “subject” of consciousness is an embodied self. Without bodily sensations, the self would cease to exist (Sentio, ergo sum).
Embodied cognition is not without unresolved issues. Whether embodiment is something that is reducible to neural representations, or requires some forms of complex coupling between brain, body, and environment, is one of the central issues that defines debates about cognition (Newen, De Bruin, & Gallagher, 2018).
As originally introduced, the term “enaction” was used to signify the lived experience of voluntary action (Varela, 1996; Varela, Thompson, & Bosch, 1991). As used today, enaction is the idea that organisms create their own experience through their actions; it assumes that organisms do not passively receive information from their environments but are actors such that what they experience is shaped by how they act (Hutchins, 2010). Organisms participate in the generation of meaning through their bodies and action, often engaging in transformational and not merely informational interactions: They enact their world (Di Paolo et al., 2010).
Drawing on a combination of neurobiological resources and phenomenological insights, Varela et al. (1991) proposed an enactivist account of cognition that emphasizes the role of a dynamic coupling of brain, body, and environment (Newen et al., 2018). Enactive models locate conscious phenomenology in the dynamic interaction of organisms with the external world (Velmans, 2007). This enactive approach posits a “deep continuity between mind and life,” where cognitive processes are seen in terms of how living systems enact relationships with the environment that promote their survival and well-being (van der Schyff & Schiavio, 2017).
O'Regan and Noë (2001) proposed that seeing is a way of acting—a particular way of exploring the environment. Perceptual “filling in,” whereby perception presents a unified mental image to consciousness in the absence of objective sensations for the totality of the image, provides a deeper meaning to the concept of enaction (Pessoa, Thompson, & Noë, 1998). A similar case is made by the stories of people who can navigate the world and communicate with others despite losing critical senses and abilities because of neurological damage (Sacks, 2010). In both cases, the perceived world is more than a set of coordinates corresponding only to what the senses detect of the objective reality of the external world, but a subjective and unified percept created (enacted) through the perceptual capacities of the organism.
Numerous authors see brain function to be first and foremost a controller of the agent's body in a meaningful interaction with its environment (Anderson, 2014). For Buzsáki (2019), this means that the brain is a self-organized system whose main job is to generate actions and to examine and predict the consequences of those actions. Clark (2016) sees the brain as an instrument evolved to anticipate the incoming streams of sensory stimulation before they arrive. This predictive brain “is not an insulated inference engine so much as an action-oriented engagement machine.”
Enactivism also has affected views of consciousness. For Hanna and Thompson (2003), enactive minds comprise conscious processes (sensory, perceptual, imaginational, emotional-affective, and volitional) that are fully integrated with the self-organizing dynamics of the neurobehavioral processes of animal bodies, both of which in turn are fully embedded in and in constant interaction with their external environments. As Noë (2010) says, “Consciousness is not something that happens inside us. It is something we do or make. Better: it is something we achieve.” In addition, for some, consciousness is a dimension of living forms that move; a creature's corporeal consciousness is first and foremost a consciousness attuned to the movement and rest of its own body (Sheets-Johnstone, 1999). Movement, thus, is the condition for a living body to interact with a meaningful milieu.
All the perspectives discussed to this point share, in varying degrees of overlap, concepts from the philosophy of phenomenology, which emphasizes the experiential character of perception and mind. The methodological difficulty of objectifying what are essentially first-person, subjective experiences kept classic cognitive science and phenomenological approaches apart (Searle, 1992). Furthermore, as Anglo-American philosophy, contrary to the Continental trend, became increasingly hyperanalytical and semantic (Schwartz, 2012), doubt that the “hard problem” of subjective experience (Chalmers, 1995) could have any physical explanation became widespread among philosophers, and biologists tended to set aside or circumvent any philosophical issues related to the mind–body problem (Crick & Koch, 1998).
Toward the end of the previous century, however, Gallagher (1997) insisted that strict lines cannot be drawn between third-person theory and first-person phenomenological description and that the division of labor between phenomenology and cognitive science is not as strict as might have been supposed. In recent years, the relevance of phenomenology to cognitive science has been revived by a heightened interest in phenomenal consciousness, by the advent of the embodied and enactive approaches described above, and by marked advances in neuroscience (Gallagher & Zahavi, 2008).
A deliberate attempt to marry the concepts of phenomenology with the methodological approaches of cognitive science was spelled out by Varela (1996). The working hypothesis of this framework, which he called “neurophenomenology,” was that the structure of experience and its counterparts in cognitive science relate to each other through reciprocal constraints. Phenomenological accounts provide information on the firsthand qualities of experience, and structural accounts provide constraints on empirical observations about the correlates of that experience. Although subjective experience is clearly a personal event, that does not make it a private event. Disciplined first-person accounts can and should be an integral element of the validation of any presumptive neurobiological corollary of the experience.
Neurophenomenology implies that consciousness and a physical substrate can at least be linked, if not superimposed. In the view of Silberstein and Chemero (2012), the hard problem of qualitative experience is transformed into the purely empirical problem of explaining how extended phenomenology–cognition works. Although still a hard problem, they see it as a tractable, empirical one. Conscious experience need not be reified in the form of qualia nor deflated to the point of reduction.
Phenomenologists maintain that a prerequisite for addressing the hard problem of consciousness is a clear understanding of what consciousness is in the first place. The goal of neurophenomenology has been to marshal the explanatory resources of both phenomenology and cognitive science to yield a coherent answer to this preliminary question. As Gallagher (1997) has suggested, the gap that persists between phenomenology and cognitive science may still seem mysterious to some, but neurophenomenology offers the potential to bring them into greater complementarity and thereby help to dissipate the mystery.
In arguing for a turn away from classic cognitive science toward embodied cognition, Clark (1997a) advocated the use of new tools, such as dynamic systems theory, which challenged cognitive science to look harder at temporally extended processes that span brain, body, and world.
Engel (2010) advocated turning the whole theory of cognition into a theory of action. System states acquire meaning by their relevance in the context of action. For this author, the intrinsic dynamics of the brain are very important, and its neural states are seen as providing the capacity to structure situations through action, rather than “encoding” information about pregiven objects or events.
The skilled intentionality framework, defined as “the selective engagement with multiple affordances simultaneously in a concrete situation,” more specifically addresses the dynamical relations of brain and body with a “landscape of affordances” (Rietveld, Denys, & Van Westen, 2018). This form of neurodynamism proposes to bring together the embodied/enactive cognitive framework (Silberstein & Chemero, 2012) with the ecological theory of affordances developed by Gibson (1977).
In another sense of neurodynamism, several authors have emphasized the importance of movement in cognition and consciousness. David Morris (2004) argues that the body schema should in fact be better conceptualized as a “moving schema of perception” that binds together a moving body and its world. It is not by our mere presence but by our moving about in the world that we are able to participate in the tactile world, and this participation is the condition for our experience of it. Buzsaki and Llinas (2017) noted that establishing a relationship between neuronal activity and distances in the world requires ambulation. Merker (2005) suggested that consciousness arose as a solution to problems in the logistics of decision-making in mobile animals with centralized brains. Sheets-Johnstone (1999) even offers the radical view that, wherever there is an animate form, and thus a moving interaction with an environment, there is consciousness.
Neurodynamism is a catch-all term for various ideas bearing on cognitive neuroscience. Perhaps noteworthy is the fact that machine learning and computation do not (necessarily) occur in moving bodies and so are dynamic only in an internal, nonextended sense. This keeps them mired in nonliving entities.
Models and concepts often overlapping with but distinct from those detailed above have also contributed to the maturing landscape of cognitive neuroscience. Several will be mentioned briefly.
Integrated information theory (IIT) was proposed by Tononi (2012) to account for empirical findings about the neural substrate of consciousness. IIT starts with the experience itself, which is intrinsic, personal (existing only for individuals in which it occurs), structured, unified, and specific (Koch, 2018). The IIT of consciousness deduces from the properties of phenomenal experience those properties and capabilities that a physical substrate of consciousness would need to possess (Tononi, Boly, Massimini, & Koch, 2016).
“Neurobiological naturalism” is the term that Feinberg (2012) coined for his version of “weakly emergent nonreductive physicalism.” It consists of three postulates (Feinberg & Mallatt, 2016). First, sensory consciousness can be explained by known neurobiological principles. Mental phenomena are caused by neurophysiological processing in the brain and are themselves features of the brain. Second, sensory consciousness is ancient and widespread in the animal kingdom, and diverse neural architectures can create it. No single emergent process explains all of sensory consciousness. Different forms of consciousness have evolved independently, and different levels of awareness have emerged gradually over evolutionary time. Third, the philosophical issues of ontological subjectivity, neuro-ontological irreducibility, and the “hard problem” can be explained by the nondissociable confluence of neurobiological and adaptive neuroevolutionary events. Ultimately, the function of consciousness is adaptive. The purpose of qualia, for instance, is to give animal-graded features of sensory input, providing more detailed and nuanced information about its environment (Edelman, 2003; Feinberg, 2001).
Neurobiological naturalism seeks to bridge and naturalize the ontological gap between subjective experience and normal biological science. Although clearly in tune with much of standard cognitive science, Feinberg and Mallatt (2016) argue that a satisfying and complete explanation of primary consciousness requires a confluence of points of view, necessarily including neurobiological, evolutionary, and philosophical arguments.
Tononi and Edelman (1998) proposed a dynamic thalamocortical core hypothesis about the properties of the neural substrate of consciousness. Their idea centers around the notion of reentry or ongoing recursive signaling across multiple reciprocally connected brain regions present mainly in the thalamocortical system (Edelman, 2001). Consciousness arises as a result of integration of many inputs by reentrant interactions in the dynamic thalamocortical core (Edelman, 2003).
The global neuronal workspace framework postulates that, at any given time, many modular cerebral networks are active in parallel and process information in an unconscious manner (Dehaene & Naccache, 2001). Information becomes conscious, however, if the neural population that represents it is amplified by an attentional mechanism into a brain-scale state of coherent activity that involves many neurons distributed throughout the brain (Dehaene & Changeux, 2011). The global availability of information through this workspace is what is subjectively experienced as a conscious state.
The dynamic core and global workspace hypotheses were independently put forward to provide mechanistic and biologically plausible accounts of how brains generate conscious mental content. The dynamic core proposes that reentrant neural activity in the thalamocortical system gives rise to conscious experience. Global workspace mechanisms reconcile the limited capacity of momentary conscious content with the vast repertoire of long-term memory. In Edelman, Gally, and Baars (2011), this relationship allows for a strictly biological account of phenomenal experience. We would note, however, that phenomenologically informed approaches to consciousness render the dynamic core and global workspace hypotheses, singly or in combination, as incomplete explanations of consciousness, for they ignore the roles of body and world in the constitution of experience. At the very least, the range of relevant information should be understood to incorporate the situation of the organism in its surroundings.
NEUROSCIENTIFIC SUPPORT FOR DIFFERENT PARADIGMS
Although a large body of experimental data in neurophysiology, neuroanatomy, animal behavior, and psychology fits comfortably within the framework of classic cognitive science, the turn toward embodiment and embeddedness in the newer strains of contemporary cognitive neuroscience further illuminates a substantial body of empirical observations. Both cases will now be illustrated, with a focus on ethology, neuroanatomical topology, and the issue of representationalism—all of which illustrate the central role that place holds in the cognitive life of the organism.
The emphasis that Von Uexküll (1926) placed on the species-specific, spatiotemporal, “self-in-world” reference frame in which organisms live their lives set the stage for Gibson's (1979) ecological psychology. Ethologist and comparative animal psychologists have echoed this perspective. William Hodos (1986) has been a prominent exemplar, arguing that natural selection optimizes mechanisms of perception maximally appropriate for the ecological demands of each species. According to Hanna and Thompson (2003), the minds of sentient organisms are fully embedded in and in constant interaction with their external environments.
Lefebvre and Sol (2008) pointed out that ecological principles like the unpredictability of resources in space and time may drive different types of cognition (e.g., social and nonsocial) in the same direction. Mettke-Hofmann (2014) identified three major environmental factors that interact with cognition: environmental variation, environmental complexity, and predation. Rosati (2017) reviewed evidence that “foraging cognition”—skills used to exploit food resources, including spatial memory, decision-making, and inhibitory control—varies adaptively across primates. Several authors have viewed cognitive processes in terms of how living systems relate to their environment in ways that are relevant to their survival and well-being, highlighting the ecologically situated nature of living agents (Fuchs, 2018; van der Schyff & Schiavio, 2017; Willemet, 2013).
Although ecological similarities may drive different types of cognition in the same direction, similar cognitive trends can be seen in taxa that are phylogenetically distant and possess remarkably different brains (Lefebvre & Sol, 2008), suggesting the possibility of convergent cognitive processes with different neural architectures, in agreement with Feinberg and Mallatt (2016).
Neuroanatomical Topology and Mapping
In the late 19th century, neuroanatomists and neurophysiologists began to amass information that tied the functional organization of the brain explicitly to the body, providing a physical substrate for the embodiment of experience.
As early as 1870, Gustav Fritsch and Eduard Hitzig had shown that electrical stimulation of specific sites in the frontal cortex of dogs evokes movement of different parts of the body, demonstrating localized motor control in the cerebrum (Fritsch & Hitzig, 1870). At midcentury, a flurry of discoveries revealed further evidence of a direct correspondence between brain organization and bodily anatomy. Penfield and Rasmussen (1952) demonstrated that motor control by the cerebral cortex is organized topologically (contiguous sites in the brain control movement in contiguous parts of the body) and that the amount of cortical tissue devoted to a body region is proportional to the precision and detail of that region's motor control. Then, Penfield and Herbert Jasper demonstrated that cutaneous sensation projects to topologically ordered sites in the sensory cortex and that the amount of cortical tissue devoted to a region of the body's surface is proportional to the precision of sensation at that point (Penfield & Jasper, 1954). Experimental findings from different species have shown that most, if not all, of the mammalian parietal cortex is devoted to somatosensory functions, which are typically, though sometimes discontinuously, represented topologically (Kaas, 1999).
Although these findings of an orderly arrangement of loci in the brain corresponding to the body periphery and interior echo the standard representational view of the world as projected into the brain, that standard view understates the degree to which the brain is organized principally to mediate an embodied interaction between the animal and its environment. This is seen especially in the way the brain allocates neural mass according to the precision of muscle control or sensory definition specific to the needs of each species, rather than a veridical map of the body and its environment.
Theoretical speculation that the brain encodes a map of its environment can be traced to Tolman (1948), who hypothesized that rats, in learning to run a maze, form a cognitive map of their environment, enabling them to situate themselves in place relative to their surroundings. The image conjured by this concept is that of the brain as a passive screen on which information from the environment is neutrally projected. In the discussion to follow, we argue that the brain is more dynamically engaged in the animal's embodied orientation and emplacement in an environment that it has enacted.
Empirical support for the existence of cognitive maps emerged essentially in three phases (Moser, Moser, & McNaughton, 2017; Moser & Moser, 2016). In the first phase, John O'Keefe and colleagues discovered “place cells” in the hippocampus that fire when a rat is exploring a particular location in its environment, irrespective of motivation or sensory modality (O'Keefe & Nadel, 1978; O'Keefe, 1976; O'Keefe & Dostrovsky, 1971). This provided strong empirical support for the neurophysiological representation of cognitive maps (O'Keefe & Nadel, 1978; O'Keefe, 1976). The second phase started with the discovery of head direction cells and the realization of the importance of angular and linear movement integration in generating spatial maps (Finkelstein, Las, & Ulanovsky, 2016). A third epoch began when investigators turned their attention to the entorhinal cortex, which led to the discovery of grid cells and border cells. Subsequent studies revealed a set of functionally dedicated cell types, namely, grid cells, border cells, head direction cells, and speed cells (Rowland, Roudi, Moser, & Moser, 2016), which act together to enable an animal to find its way from place to place. Evidence for grid-cell-like representations in humans has been detected as well (Doeller, Barry, & Burgess, 2010). The cortical circuit for spatial representation thus has multiple functionally distinct components, each dedicated to a highly specific aspect of spatial processing (Stensola & Moser, 2016).
It should be noted that functions we interpret as map-like representations in the brain may alternatively be characterized as structured sequences of neuronal cell assemblies whose function is to infer trajectories through the surroundings that the animal lives in and explores (Buzsaki & Llinas, 2017). Certain hippocampal cells are thought to abstract spatial concepts from the superficial details of the environment and encode space into a schema-like representation (Baraduc, Duhamel, & Wirth, 2019). Rather than computing spatial locations, the space that hippocampal cells encode may better be described as a life space that encodes the history of experience into the relational structure of episodes (Shapiro & Eichenbaum, 1999).
The hippocampal–entorhinal cortical system is clearly concerned with an animal's orientation toward its environment or the way in which it is embedded within it. However, just as clearly, this system is concerned with more than simple location (Cooper & Ritchey, 2020; Devenport, Hale, & Stidham, 1988).
The existence of place, grid, border, and head direction cells in the hippocampal–entorhinal complex and transmission of the information they mediate to higher-order processing sites illustrate the neural mechanisms that are necessary for an organism to participate in its environment. However, a simple correspondence between external location and internal topology is insufficient to account for how these neural maps promote the adaptive interests of organisms for their survival and enact behavior that facilitates the navigation of a life or problem space (Shapiro & Eichenbaum, 1999) through a multimodal, situational survey of the surrounding world and body (Pennartz, Farisco, & Evers, 2019). We argue below that this can best be understood in terms of dynamic inhabitation of place.
Why is the mind in the head? McCulloch (1951) answered, “Because there, and only there, are hosts of possible connections to be formed as time and circumstance demand.” In other words, as viewed in classic cognitive science, the processes that govern animal behavior rely on inputs from a host of external and internal sources that need to be integrated and processed for generating an appropriate output. This necessarily depends on a single site of integration and organization (the brain), where both the sources of input and targets of output can be epitomized for analysis (Churchland, 2002).
Despite the common-sense logic that an animal has to be able to internalize the world in which it lives to some degree to behave appropriately within it, the concept of internal representation (in the brain) has fallen out of favor among many contemporary philosophers and some cognitive scientists, who see cognition as an extended process that encompasses the whole body and environment outside the head in a coordinated brain–body–environment system (Engel, 2010; Ennen, 2003; Clark, 1997b; Gallagher, 1997; Clark & Toribio, 1994). The ultimate consequence of such a full-strength “embodied, embedded, extended, and enactive” revolution is a commitment to “externalism” as well as antirepresentationalism (Wheeler, 2017; Velmans, 2007). Rhetorical support for this framework outpaced both evidence and argument, in the view of Clark and Toribio (1994), who nonetheless saw the antirepresentationalist challenge as having the virtue of forcing consideration beyond the bare representational/nonrepresentational dichotomy to awareness instead of a rich continuum of degrees and types of representation.
As Clark (1997a) has pointed out, “representationalism” has been used with four different meanings. In the first and weakest sense, it refers to the bare idea of an internal state of some sort in the brain, which virtually no one denies. In a second and stronger sense, it means an internal state correlated with aspects of the body's internal or external environment. A third, more explicit and controversial meaning is that of an inner state that functionally represents (in a cause–effect way) other objects, events, actions, or states of affairs. The fourth use of the term is an expansion of the third definition, extended to include off-line processing.
Many examples of evidence for representationalism of the third type can be cited. The following is a sample, far from comprehensive. Much of the forebrain, especially the dorsal thalamus and neocortex, consists of nuclei and areas that are parts of complex systems that analyze sensory information and allow behavior to be guided by accurate inferences about the external world (Kaas, 1989). Specific hippocampal cells encode information about the perceptual and behavioral structure of experiences (Shapiro & Eichenbaum, 1999). Recent findings have shown that the hippocampal memory trace serves as an index for a cortical representation of memory (Tanaka & McHugh, 2018).
Since its inception by Donald Hebb in 1949, the concept of the “cell assembly” has been one of the most enduring and influential motifs in neuroscience. Hebb (1949) proposed a “neuropsychological” theory that thoughts, images, and qualitative experiences depend on specific patterns of neuronal activity in which sensory stimulation activates a discrete set of neurons in a specific spatiotemporal pattern. Such a cell assembly would be the unit of perception at the multicellular level in the brain. Complex perceptions lead to the association of cell assemblies into a “phase sequence.” The association of cell assemblies is made repeatable (learning) and retrievable (memory) by changes in the synaptic efficacies within specific neuronal circuits underlying a phase sequence.
Mountcastle (1978) pointed out that neuronal populations are organized in multiply replicated local neuronal circuits constituting columns oriented perpendicular to the cortical surface. These columns, in turn, are composed of closely linked subsets or minicolumns (Kaas, 2012). Combining the two concepts—cell assemblies and minicolumns—has made it possible to envision a neurophysiological substrate for a unit (or fragment) of cognitive experience at the level of specific neuronal pathways. The columnar organization of the cerebral cortex, or some similar substrate for cell assemblies, is central to representationalist theories of information in the brain, including some invoking consciousness (Dehaene, Lau, & Kouider, 2017; Hawkins, Ahmad, & Cui, 2017; Mercado, 2008; Calvin, 1996; Edelman, 1987, 1989; McClellan & Rumelhart, 1985)
Despite the attractive features of the cell assembly concept, evidence for the reality of functional cell assemblies until lately has remained largely elusive. However, significant progress recently has been made toward finding cell assemblies, especially those that underlie memory traces, often referred to as engrams. New methodologies using novel approaches to tag populations of neurons that are active during memory encoding, and further allowing these engram-associated neurons to be manipulated at later times, represent considerable progress in allowing engrams to be observed, expressed, and erased (Lacagnina et al., 2019; Pignatelli et al., 2019; Takamiya, Yuki, Hirokawa, Manabe, & Sakurai, 2019; Abdou et al., 2018; Josselyn, Kohler, & Frankland, 2015; Restivo, Niibori, Mercaldo, Josselyn, & Frankland, 2015; Tonegawa, Pignatelli, Roy, & Ryan, 2015).
Assuming that engrams consisting of phase sequences of cell assemblies unique to specific elements of experience do exist, how can they be modified over the life of the organism? Synaptic plasticity is most often assumed to be the mechanism by which specific circuits can be added to or removed from cell assemblies, leading to engram storage (learning) or loss (forgetting). The relevant mechanisms may be neurophysiological (de Sousa et al., 2019; Kastellakis & Poirazi, 2019), microanatomical (Schafer et al., 2012; Hofer, Mrsic-Flogel, Bonhoeffer, & Hubener, 2009; Yang, Pan, & Gan, 2009), neurochemical (Butler, Wilson, Mills, Gunnersen, & Murphy, 2019; Matos et al., 2019), or a combination of all of these (Joshi, Patel, Rehan, & Kuppa, 2019).
The plasticity of synaptic connections has not been the only mechanism envisioned for changes in cell assemblies over time. One example proposed a tripartite mechanism involving the interactions of neurons with their neural extracellular matrix, trace metals, and neurotransmitters as the basis for a biochemical memory engram (Marx & Gilon, 2019). Decades ago, Galambos (1961) proposed that glial cells could act to organize neuronal circuitry. For years, Ross Adey (1969, 1988) advocated a tricompartmental model in which neuronal, neuroglial, and extracellular compartments would constitute a micrometabolic module for affecting local brain cell dynamics.
In one sense, the examples above provide evidence for the reality of representationalism, in Clark's (1997a) third and fourth uses of the term, and demonstrate numerous plausible mechanisms for plasticity in those representations. However, we note that this very plasticity argues against a conception of cognitive functioning as operating on a passive mirror of the external world, because interaction with the world actively alters neuronal structures. Copious evidence from across the cognitive sciences demonstrates that such altered structures shape an animal's perceptions and behaviors. Cognitive ecology; “embodied, embedded, extended, and enactive” cognition; neurophenomenology; and other frameworks increasingly influential in contemporary cognitive neuroscience can help to provide the conceptual resources for understanding this dynamically recursive interaction. To the extent that “representation” is retained as a useful term in cognitive neuroscience, it is best used to indicate this dynamic process. To the extent that it distracts from this process and tends to promote an image of the brain as passively imprinted by an objective world, its use is counterproductive.
ARGUMENT FOR THE PRIMACY OF PLACE
Animals have evolved in constant and intimate interaction with their environments. They have been embedded from the beginning. Their very physiognomy has been shaped by the nature and constraints of their surroundings. Contemporary cognitive neuroscience is showing a growing appreciation for the view that animals thus enact a dynamic interaction with what their environments afford and their bodies make possible.
As reviewed above, cognitive ecology, embodied cognition, enaction, neurophenomenology, and various other frameworks have become parts of the contemporary landscape of cognitive neuroscience. All these newer frameworks share, with varying degrees of emphasis, the view that cognition is embodied, extended, embedded, and enacted. They also share, either explicitly or implicitly, a reliance on “place,” which we think deserves greater focus and a deeper understanding than it is usually accorded. We believe, in fact, that the time has come for neuroscience to develop a more theoretical and coherent concept of place.
Traditionally, cognitive research has assumed that the brain orients the animal in space according to an abstracted representation of the world in terms of Cartesian coordinates external to the animal and has sought to discover what the neuroanatomical and neurophysiological substrates of that representation are.
However, that is not the way that animals experience place. We contend that place, from the viewpoint of an embodied animal, is best understood as “the milieu in which the animal is situated, consisting of the affordances the milieu provides of relevance to it.” In other words, the animal enacts a species-specific perception of place, in terms and details that are meaningful to that particular animal's existence in that specific situation. A similar definition in terms of humans specifically defines places as “spatial fields that gather, activate, sustain, identify, and interconnect things, human beings, experiences, meanings, and events” (Seamon, 2018). We would add that to be emplaced in this way is the condition for all experience.
Viewed in this light, the concept of place is central to a number of cognitive functions, listed as follows with brief explanations.
The Interdependence of Place, Space, and Time
The argument has been made that space and time are not preexisting categories, but rather belong to a primordial unity (Buzsaki & Llinas, 2017), and furthermore that distance and duration are not derived from first principles (Buzsáki, 2019). Instead, vision, hearing, olfaction, and proprioception are used to deduce the location of and distances to objects. As the animal moves through the environment, the map is formed by multiple mechanisms, including counting the number of steps, self-motion-dependent optic and somatosensory signals, and vestibular acceleration signals. The animal's velocity converts between distance and duration representations, making place cells and time cells equivalent. Everything that we attribute to time in the brain can be accomplished by sequential cell assemblies or neuronal trajectories. (Buzsáki, 2019; Buzsaki & Llinas, 2017; Bonato, Zorzi, & Umiltà, 2012). Time derives from a perception of space, and space is determined by movement through a succession of places.
These observations rightly suggest that the concepts of abstract space and time do not exist independent of embodied experience, nor can they be understood as independent of each other. We propose that, at least for sentient organisms, perceptions of space and time derive from a more fundamental cognitive feature—the perception of place. The mere presence of an animal in an environment gives it no information on the dimensions of its surroundings. Rather, spatiality and temporality arise fundamentally through action; engagement of the animal's active body with its world is the origin of its perception of space (Merleau-Ponty, 1945, p. 262).
Detection of Place and Orientation
For Merleau-Ponty (1945), the perceived world is grasped only through orientation, hence being cannot be dissociated from oriented being, and lived space is always oriented space. Control of appropriate actions requires information about the spatial relationships among objects and surfaces in the world, relative to the orientation and configuration of the animal's body (Cisek, 2007).
Evidence for neural substrates that underlie an animal's ability to orient itself in space has grown in recent years. The discovery of multiple, clustered place fields with correlated movement-tuning properties in small neural ensembles in the hippocampus suggested the existence of neuronal networks serving to encode multiple sensory and behavioral aspects of a place or event (Eichenbaum, Wiener, Shapiro, & Cohen, 1989). Animal studies provided further evidence that vector-based representations are employed in spatial cognition (O'Keefe, 2003). Cortical circuits for spatial representation consist of multiple functionally distinct components, each dedicated to a highly specific aspect of spatial processing. As reviewed in an earlier section, the circuits include place cells in the hippocampus as well as grid cells, head direction cells, and border cells in the medial entorhinal cortex (Moser et al., 2017; Stensola & Moser, 2016; Doeller et al., 2010; Hafting, Fyhn, Molden, Moser, & Moser, 2005).
The Way an Animal Enacts Its Lived Space
Lived space emerges out of the interaction between an active body and its environment. There is no bodily action that is not an engagement with the body's surroundings. If all experience is embodied, then the meanings that an animal might find in a place will be determined, on the one hand, by its bodily capacities and, on the other hand, by the affordances of its milieu that solicit those capacities into action. There is thus a reciprocal relation between body and place that is the condition for all experience.
The Role That Place Plays in Memory Encoding and Retrieval
One of the major achievements of cognitive neuroscience over the last three decades has been discovery of the existence of brain circuitry that enables an animal to be situated and oriented in a place-specific manner within the milieu it inhabits. However, the early emphasis on a narrow interpretation of place, simply as a punctate location in a Cartesian matrix, has been revealed to be overly simplistic. It has turned out that these circuits are also intimately connected to the formation of new memories (Moser & Moser, 2016). As Eichenbaum et al. (1989) argued, hippocampal processing is not limited to the representation of spatial location. Rather, multiple cues activate specific cells that mediate numerous sensory and behavioral aspects of a place or event (Wiener, Paul, & Eichenbaum, 1989). Thus, the hippocampus perhaps more accurately provides a memory map than a spatial map (Shapiro & Eichenbaum, 1999). Or, perhaps the encoding of memory is carried out in the brain as a form of placement.
Over 30 years ago, Teyler and DiScenna (1985) proposed a memory indexing theory, in which experiential events are initially stored in an index of neocortical locations maintained in the hippocampus. Subsequently, all the different neocortical regions mediating different elements of the experience permanently encode them and their interrelationships. More recently, Tanaka et al. (2018) made a similar suggestion. In his version, the hippocampal memory trace serves as an index for a cortical representation of memory (a map for internal representation), with the primary hippocampal function being to reinstate the pattern of cortical activity present during encoding (Tanaka & McHugh, 2018).
Spatial memory appears to be supported by multiple parallel patterns of neural activity (Burgess, 2008). Information possessing the feature of situatedness, in combination with memory capacities, enables the formation of episodic memory, defined via the conscious recall of occurrences specified by event identity, place, and time (Pennartz et al., 2019). All these observations implicating a role for neural substrates long associated with the specification of place in the encoding of memory constitute growing evidence for an interdependence between memory and place, consistent with long-held anecdotal convictions.
The Role of Place in Thought and Consciousness
Every thought is a bodily experience. If all experience is emplaced, then all thought must be emplaced. A conscious agent makes and remakes a meaningful world from a succession of things and places that comprise its experience. If thought is an expression of an embodied experience of place, and if there are many ways to inhabit places, then the way that any species inhabits the world will depend on how that species employs its unique bodily capacities. In other words, styles of thought will be expressed as styles of embodied engagement in place.
Several lines of analysis have suggested ways that abstract thought processes may be grounded in embodied and emplaced experience (Stocker, 2014, 2016; Woelert, 2011; O'Keefe, 2003). Conceptual metaphor theory has developed a framework for considering abstract thought as originating in bodily experience, going back to Lakoff and Johnson (1980). Implicit in the concept of cognition as extended brain, body, and niche (Silberstein & Chemero, 2012) is the assumption that they are situated in a place at any given point in time.
In this review, we have covered the major ways in which new frameworks have been superimposed upon and, in some cases, have challenged the standard assumptions of classic cognitive science. These have become sufficiently influential to constitute what we view as a newer set point for contemporary cognitive neuroscience. In doing so, we have not meant to question the value of research based on the classic models or deny the validity of insights gained through the perspective of those models. What we do suggest is that the contemporary state of cognitive neuroscience now calls for some adjustments to experimental strategies and objectives. Several of the most important ones will briefly be touched upon.
As evidence grows for the ancient origins and widely differing mechanisms of animal cognition (Irwin, 2020), the need to conduct more comparative studies with a greater variety of species becomes clear. For example, the highly diverse behavioral repertoire of reptiles, with their paleocortical precurser of the mammalian hippocampus, illustrates the value of a comparative approach toward understanding hippocampal function. (Reiter, Liaw, Yamawaki, Naumann, & Laurent, 2017).
The argument that sensory consciousness is ancient and widespread in the animal kingdom, and that diverse neural architectures can create it (Feinberg & Mallatt, 2016), raises the question of which cognitive processes and underlying physiological/neuroanatomical states have deep roots because of needs shared by all animals and which are differentiated and derived because of environmentally specific adaptations.
The situated and embedded nature of an animal's natural cognitive milieu requires greater attention. Neuroanatomical complexity and cognitive abilities are correlated with environmental uniqueness and variability (Rosati, 2017; Mettke-Hofmann, 2014; Shumway, 2008), so experimenters should conduct more studies of cognition in natural, or at least more complex, environments.
If, as Merleau-Ponty (1945) wrote, “A sense of space emerges through movement within a milieu,” and movement is a precursor for consciousness (Sheets-Johnstone, 1999), the contribution of movement to animal cognition needs more emphasis and exploration. Merker (2007) has argued that consciousness arose as a solution to problems in the logistics of decision-making in mobile animals. What the cognitive systems for such decision-making are deserves more study.
Recognition that behaving animals internalize a concept of place that goes beyond simple localization calls for new and creative experimental approaches. The well-documented mechanisms in the hippocampal–entorhinal cortical system for placement and orientation of mammals is almost certainly an incomplete picture of the entire neural substrate for how the embodied animal senses and generates spatial information about its environment. How that system incorporates multimodal sensory information, enacts motor activity, and integrates information at higher levels of integration needs further flushing out.
The confounded use of “representationalism” in cognitive science needs to be untangled. The clear evidence from neuroscience that elements of the extended brain–body–environment are correlated with defined neural processes does not negate the insight that neural representation, although in a certain sense necessary, provides an incomplete understanding of cognitive functions and consciousness. Clark (1997a) argued that putative internal representations may involve indexical or action-oriented contents rather than compositional codes. In a similar vein, Gallagher (2014) warns against a version of embodied cognition that leaves the body out of it—placing instead all the essential action of cognition in the brain. The content of consciousness should be construed as a dynamically recursive interaction between a whole organism and its milieu.
The ability to visualize engrammatic traces of memory in the brain is one of the most exciting recent advances in neuroscience, but this technology is based excessively on one species (the mouse) and one learning paradigm (contextual fear conditioning). Until a greater variety of animals and behaviors are investigated, generalizations from currently available data need to be tempered.
The neurophenomenological approach advocated by Varela (1996), using first-person reports to guide neural analyses in the study of subtle human consciousness states (Berkovich-Ohana, 2017), should be an integral element of the validation of neurobiological processes in humans. Phenomenological descriptions can influence the design of scientific experiments to help neuroscience determine more precisely what phenomena it should explain (Gallagher & Zahavi, 2008). As Crick, Koch, Kreiman, and Fried (2004) pointed out, neurosurgeons, probing the living human brain on a daily basis, could make significant contributions in this regard.
For those neuroscientists willing to tackle the “hard problem” of reconciling subjective experience with objective physical substrates (Chalmers, 1995), we suggest that attention to the embodied, embedded, and interactional nature of consciousness is more likely to result in a fruitful enquiry. We do not claim that any of the new approaches in cognitive neuroscience have solved that problem, nor that they necessarily will, but we do maintain that these approaches go further than classical cognitive science in properly addressing the preliminary question posed by phenomenology, namely, what is the nature of consciousness in the first place? To this question, we contribute the suggestion: It is always an occurrence in and of place.
The distinctive features of consciousness need to be defined in neurocognitive detail. The neuroanatomical substrate of awareness in the mammalian brain is fairly well understood. Further studies have suggested some previously unrecognized but specific neurophysiological correlates that may distinguish brain activity tied to conscious awareness (Massimini, Boly, Casali, Rosanova, & Tononi, 2009; Seth, Baars, & Edelman, 2005). Other approaches, such as IIT, which deduces from the essential properties of phenomenal experience, the requirements for a physical theory of consciousness (Tononi et al., 2016), might be a productive framework for further investigation, provided it is expanded to incorporate environmental interaction.
Feinberg (2012) has argued that a fuller understanding of consciousness will require a combination of evolutionary, neurobiological, and philosophical approaches (Feinberg & Mallatt, 2016). In advocating for a more phenomenological approach to the study of consciousness, Gallagher has written, “cognitive science [and] phenomenology [both] view consciousness as a solvable problem and affirm its openness to scientific, objective interpretation. In this case it is only the gap that continues to persist between phenomenology and cognitive science that seems mysterious.”
Many contemporary cognitive neuroscientists are heeding Andy Clark's (1997b) admonition to abandon the idea of neat dividing lines between perception, cognition, and action and to avoid research methods that artificially divorce thought from embodied action-taking.
Gallagher (2018) recently suggested that the best explanation of brain function may be found in the mixed vocabularies of embodied and situated cognition, developmental psychology, ecological psychology, applied linguistics, and the theory of affordances and material engagement, rather than the narrow vocabulary of computational neuroscience.
To these perspectives, we would add that a more intensive focus on the primacy of place will further a better understanding of the cognitive life of humans and other animals, in a manner specific to the unique needs of each.
Reprint requests should be sent to Louis N. Irwin, Biological Sciences, University of Texas at El Paso, or via e-mail: email@example.com.