Imagination As A Fundamental Function Of The Hippocampus Part 1

Imagination is a biological function that is vital to human experience and advanced cognition. Despite this importance, it remains unknown how imagination is realized in the brain. Substantial research focusing on the hippocampus, a brain structure traditionally linked to memory, indicates that firing patterns in spatially tuned neurons can represent previous and upcoming paths in space. 

Imagination and memory are among human beings' most valuable intellectual resources. These two abilities complement and promote each other, and at the same time they also profoundly affect our lives.

First, imagination is the key to realizing our latent potential. Through imagination, we can face a variety of situations and explore new ideas and perspectives. Imagination allows us to see the world from a creative perspective and discover new opportunities and resources. At the same time, good memory also enables us to dig out important information from memory and enable imagination to be expressed more accurately.

Secondly, imagination is the soul of human innovation. Imagination helps us discover new knowledge, create new works of art, and optimize existing lifestyles. By understanding information from the past and present, we can construct an ideal future in our minds, keep our focus on the future, and constantly look for the best way to realize our desires.

Finally, imagination and memory can also improve our mental health. These two abilities allow us to better handle the various challenges and stresses in life. At the same time, the virtual experience provided by imagination can also help relieve anxiety and stress, adjust emotions, and allow us to face various changes in life with a more positive attitude.

In short, imagination and memory are interdependent. Only by strengthening the connection between the two and combining creative ideas with practical knowledge can we achieve better results. Therefore, we should work on improving our imagination and memory abilities to inject more wisdom and possibilities into our future lives. It can be seen that we need to improve memory, and Cistanche deserticola can significantly improve memory, because Cistanche deserticola can also regulate the balance of neurotransmitters, such as increasing the levels of acetylcholine and growth factors. These substances are very important for memory and learning. In addition, Meat can also improve blood flow and promote oxygen delivery, which can ensure that the brain receives sufficient nutrients and energy, thereby improving brain vitality and endurance.

Click know supplements to improve memory

This work has generally been interpreted under standard views that the hippocampus implements cognitive abilities primarily related to experience, whether in the past (e.g. recollection, consolidation), present (e.g. spatial mapping), or future (e.g. planning). However, relatively recent findings in rodents identify robust patterns of hippocampal firing corresponding to a variety of alternatives to experience, in many cases without overt reference to the past, present, or future. 

Given these findings, and others on hippocampal contributions to human imagination, we suggest that a fundamental function of the hippocampus is to generate a wealth of hypothetical experiences and thoughts. Under this view, traditional accounts of hippocampal function in episodic memory and spatial navigation can be understood as particular applications of a more general system for imagination. This view also suggests that the hippocampus contributes to a wider range of cognitive abilities than previously thought.

This article is part of the theme issue ‘Thinking about possibilities: mechanisms, ontogeny, functions, and phylogeny.

1. Introduction

The ability to imagine is essential to human experience. At a broad level, imagination has a major role in human creativity, agency, and everyday thoughts and actions. More specifically, humans have and express many types of imagined experiences. These include recollections, predictions, simulations, counterfactuals, fantasies, suppositions, and mind-wandering—and, in pathological cases, hallucinations and confabulations. 

These wide-ranging forms of imagination are relevant, if not essential, to a similarly wide range of cognitive domains, such as memory, planning, learning, and inference. Despite this fundamental importance, our understanding of how imagination is realized as a biological process in the brain remains nascent. Indeed, the sheer diversity of imagined experiences makes it challenging to begin to envision a possible biological approach.

As a starting point, we identify a unifying characteristic of imagined experiences: they do not refer to actual present experiences or directly reflect ongoing circumstances in the external world. Rather, imagined experiences refer to non-actualities, and arise from a source internal to the subject. Awake healthy subjects can, in other words, ‘mentally’ self-generate thoughts and experiences and distinguish them from thoughts and experiences driven by ongoing stimuli in the actual present. 

We refer to this fundamental ability to generate possibilities that do not correspond to the actual present as generativity. By this definition, generativity is a basic function that underlies imaginative abilities broadly, regardless of more specific properties, such as references in time (e.g. remembering the past or simulating futures).

 As further clarification, we also note that our present use of ‘generativity’ differs from its senses in linguistics and statistical models (notwithstanding potential connections between these uses [1–3]). Defining generativity enables us to focus on a single characteristic ability that may ultimately facilitate our understanding of the diverse types and components of imagination.

Crucially, generativity can be understood at the level of the brain. Mirroring the subject-level ability to distinguish actual from imagined experience [4], specific neural processes in the healthy brain must ‘parse’ internal representations as ongoing experience (actual) versus internally generated alternative experience (imagined). Importantly, this substrate-level generativity does not presuppose features such as mental imagery, mental time travel, or conscious awareness. Indeed, defining generativity enables us to refer to the brain’s capacity to internally generate experiences that are distinguished from externally driven present experiences, without invoking these features that are associated with subjective human imagination. 

As an example, a soccer player approaching a moving ball can rapidly assess numerous dynamic ongoing events and stimuli, consider multiple possible responses, and decide on a play, all in a split second and without overt awareness of each internally represented possibility. In animals, ethologically relevant scenarios such as predation and escape make similar demands on cognition [5]. Thus, direct investigation of the brain may be essential to understand generativity.

In this review, our overall aim is to describe and advance our understanding of how generativity—an ability underlying imagination—is realized in the brain. Our review is guided by five questions: (i) where generativity might be implemented in the brain, (ii) how generative neural activity can be identified, (iii) what candidate generative neural activity patterns and representational correlates have been previously described, and (iv) how the brain can organize actual versus generative activity patterns. 

This discussion establishes that the hippocampus, a brain structure in the medial temporal lobe, is a candidate biological substrate of generativity and that patterns of hippocampal neural firing reflect generative processes by representing a diverse range of alternatives to ongoing experience. Finally, we consider (v) what these observations suggest about the biological basis of generativity and its role in cognition. 

More specifically, in light of recent findings at the level of neuronal firing patterns in rodents, in addition to brain research related to imagination in humans, we suggest that the hippocampus—often understood as a system that characteristically represents experience, whether in the past, present or anticipated future—may be better understood as a system that also represents imagined alternatives to experience.

2. The hippocampus as a locus of generativity in the brain

What structure within the brain might implement generativity? One approach to this question is to determine whether damage to specific parts of the brain causes deficits in imaginative abilities relying on generativity, including recollecting the past, envisioning the future, or constructing fictional scenarios. Notably, the earliest case studies linking imagination of the future to specific brain areas are in individuals with previously established deficits in memory of the past [6–10]. 

In one classic case, patient H.M. suffered severe amnesia after his hippocampus and adjacent medial temporal areas were surgically removed, which established the hippocampus as an important site for memory, particularly episodic memory [11,12]. Notably, while episodic memory impairments are most traditionally reported, H.M. and many other patients with hippocampal damage have since been examined and found to have severe impairments in future-oriented thinking and constructing fictional events more generally [9,13–18]. 

These findings raise the possibility that recollection of the past, anticipation of the future, and imaginative abilities more broadly may share common underlying functions as well as dependence on the hippocampus [17,18].

Complementing lesion studies, functional brain imaging has revealed activation of the hippocampus during a variety of self-reported imagined experiences that overtly differ from subjects’ actual circumstances [19–22]. 

In such studies, subjects are typically asked to imagine experiences that differ from present experiences through changes in time, space, and/or personal perspective. The hippocampus, in addition to a group of cortical areas known as the default mode network, is consistently activated during, for instance, recalling autobiographical experiences, imagining anticipated future episodes, imagining counterfactuals, mentally simulating common activities (e.g. brushing teeth), constructing fictional scenes, imagining non-actual events and stories, taking on others’ perspectives and unprompted mind-wandering [19,20,23–27]. These results highlight that the hippocampus, along with other brain regions in the default mode network, is important for the capacity to generate mental displacements from actual present circumstances, whether in time, space, personal perspective, and possibly other domains [14,17,19,28,29]. 

Thus, although the cognitive role of the hippocampus is often conceptualized about prior experience (i.e. episodic recollection, recall) or explicitly anticipated experience (i.e. planning, prospection) [30–32], the hippocampus appears to play a more general role in imaginary experience [29].

In efforts to clarify this role, studies have often probed the availability and character of mental imagery. 

Several further studies help refine the role of the hippocampus beyond the observation mentioned above that hippocampal damage is associated with deficits in vividly visualizing fictional scenes. First, patients with partial hippocampal lesions show activation of residual hippocampal tissue when tasked with imagining complex scenes [33,34]. Second, one patient with longstanding hippocampal damage found it effortful but possible to visualize single imaginary objects and simple scenes, yet could not readily imagine complex scenes in one automatic and coherent picture—instead, he built up the scenes ‘bit by bit’ [33]. Residual hippocampal tissue in this patient was not activated during these tasks as it was in control participants [33]. 

These findings suggest that the hippocampus is not strictly required for mental imagery, and therefore that the role of the hippocampus in imagination may be only indirectly related to mental imagery. The requirement of the hippocampus for readily constructing complex scenes, in particular, suggests a different basis or principle by which the hippocampus contributes to the imagination [33]; we revisit this issue in the section ‘Generativity as a function of the hippocampus’.

The above lesion and functional imaging work implicate the hippocampus as a candidate substrate for generative thinking, typically by relying on conscious verbal or behavioral reports. 

This approach is, however, limited in addressing how generative processes are implemented at a neuronal level. For example, the timing of underlying processes relative to eventual behavioral reports remains unclear. Generative processes may also unfold at timescales considerably faster than behavior, which suggests the need for complementary approaches with finer temporal resolution. 

Here animal models provide an important advantage by enabling greater access to neural firing. This potential approach, in turn, raises the question of whether animals also exhibit behaviors indicating generative thought, and if so, whether the hippocampus is also implicated, as in humans.

From work dating at least a century, it is clear that animals behave based on the memory of prior experience and conceptual insight rather than solely trial and error, instinct and presently sensed information [35–37]. 

This implies a corresponding ability to construct and use internal representations and suggests the existence of generative neural processes in animals. In the case of rats, a common model for hippocampal studies, a seminal example of behavior based on internal representations is spatial navigation. When navigating, rats can take novel paths (for instance, shortcuts to goal locations), implying an internal model enabling the ability to generate such novel courses of action [38,39]. 

Rat behavior can also appear deliberative and regretful, suggestive of internally generating representations of possibilities, including counterfactual pasts [40–42]. In service of these and other behaviors, the hippocampus is thought to be essential for using an abstract internal model, or ‘cognitive map’ that relates items, events, and features of experience [42–44]. Indeed, hippocampal damage impairs various behaviors thought to rely on abstract internal representations such as rats’ abilities to infer relationships between stimuli [45]. 

Further, hippocampal lesions impair rats’ abilities to make choices dependent on an internal model and predictions or plans made by that model [46]. These findings suggest that the hippocampus is an important locus in the rodent brain for constructing abstract mental models, which in turn could be used to generate representations of prior, new, and otherwise not presently experienced possibilities, enabling insightful behaviors.

With the hippocampus as a starting point for investigating generativity in both humans and animals, we now aim to clarify what neural firing patterns have been observed in the hippocampus and what internal representations they suggest. To do so, it is necessary to address our second question: how can generative neural activity patterns be identified?

3. Identifying neural firing patterns that are generative

Identifying neural firing patterns that may represent imagined experiences requires us first to identify neural firing that corresponds to experience. Here, we focus on studies of neural firing in the rodent hippocampus. To investigate internal representations at the level of neurons, neurobiologists have leveraged the well-established relationship between spatial location and hippocampal firing in freely moving rats [47]. Over 50 years of work have established that principal neurons in the rodent hippocampus exhibit increased firing rates when the animal is in distinct physical locations (figure 1a) [47,48]. As the rat moves through an environment, each of these ‘place cells’ consistently increases its firing rate when the animal is in the neuron’s ‘place field’ location(s) [47,48]. 

Importantly, place cell firing also varies based on numerous factors besides location [49]; for example, in linear environments, a large proportion of place cells fire more when the animal is traveling in a particular direction [50]. Therefore, at a broader level, it is important to note that a place field describes average firing over many individual runs through a location, even though there is often substantial variability in a place cell’s firing across individual runs through the same place (figure 1a).

The basic notion of a place field, along with the ubiquity of place cells in the rat hippocampus, provides a possible approach to identifying actual and generative activity at a neural level. If we take a place cell’s activity to represent its place field location, then each instance of firing by that neuron can be provisionally understood as representing that location. By this interpretation, a place cell will reliably fire when the animal is in the cell’s place field, thereby representing the animal’s actual present location.
Importantly, in certain moments, a place cell can also fire when the animal is not actually in the cell’s time-averaged place field location (figure 1a,b) [51–53]. Accordingly, these moments can be provisionally understood as times in which a representation of the place field location is internally generated, even though the animal occupies a different location at that moment.

Strikingly, place cells have been found to fire outside of their place fields in coordination with each other (figure 1b) [54,55]. During these events, the collective activity of place cells can be understood to express a representation that corresponds to locations different from the animal’s current location [52,53]. In other words, this neural firing is consistent with a generative representation; while it appears displaced from the animal’s actual state and present stimuli, it is internally coordinated across cells (figure 1b).

A variety of analysis methods have been used to investigate these generative firing events and internal spatial representations in the hippocampus [56–58]. Briefly, one approach is to model the firing of many individual place cells as their time-averaged place field locations, and then invert that model to produce an estimate of the neurally represented location at each moment in time [59–61]. Doing so enables us to infer, or decode, the animal’s moment-to-moment ‘mental location’ based on hippocampal firing patterns. Thus, by identifying periods when the decoded representation of location (or direction) differs from the animal’s actual state, we can examine periods when hippocampal activity is collectively inconsistent with a representation of experience and may instead be generative. This enables us to address our third question: what kinds of generative representations have been observed in the hippocampus?

4. Generative representations in hippocampal neural firing

Single-cell and population decoding approaches have revealed a striking variety of putative generative representations in the rat hippocampus over the past several decades [62–65]. Traditionally, these representations have been accounted for as specific episodes and abstracted experiences that are based on the past, or that anticipate experiences in the future [66,67]. Recent results, however, imply that the hippocampus also regularly represents alternatives to experience, whether in the past, present, or anticipated future [68–70]. Together, these findings suggest that the hippocampus may generate a substantially wider range of internally constructed alternatives to the animal’s experience than traditionally understood.

(a) Representations consistent with past experiences

The first reports of hippocampal activity patterns related to past experiences focused on sleep [51,54]. Firing sequences of place cells that were active during running on a maze were found to reactivate in similar sequential order during subsequent sleep, as if briefly ‘replaying’ past spatial experience [71–73]. These replays occur on the order of tens to hundreds of milliseconds, far faster than the seconds-long timescale over which the actual behavioral traversal of those locations unfolds (figure 1b) [71]. Importantly, replay events were subsequently found to occur during waking periods in which rats are behaviourally immobile, such as sitting still or eating (figure 1b) [74,75]. During wake and sleep, replay typically occurs during a burst-like hippocampal network-level activity pattern, the sharp wave-ripple (SWR), that is itself internally generated (rather than externally driven), consistent with the notion of generativity [76].

As suggested by its name, replay has been interpreted as recapitulating specific episodes of prior experience. An early observation was that after an animal ran towards and then came to rest at a reward location, a path was replayed starting at the animal location and proceeding in reverse as if retracing the path that led to the reward [75,77,78]. Replay representations not only initiate at a stationary animal’s location [74] but can also correspond to paths that start farther away from the animal within the current maze, as well as on a different maze experienced beforehand (figure 1b) [79,80]. 

These examples are evocative of the hippocampus’ long hypothesized role in cognitive functions that rely on experiences from the past, such as memory consolidation and episodic recall [65,81].

Additional findings on replay suggest a more complex picture. Unlike a rigidly recapitulative process that uniformly represents recent experiences, a replay can be enriched for previously taken paths associated with reward, paths associated with aversive outcomes, nearby locations, and paths that have not recently been taken [61,82–84]. Further, these and several additional findings [82,84–88] suggest that replay events are collectively well described as reflecting an abstract internal spatial model of the encountered environment, or a spatial ‘cognitive map’ [43,52,62]. 

For instance, replays can be biased toward paths that are less behaviourally traversed, and replays can be consistent with random trajectories through a familiar space [87,88]; replays like these may sample locations that are not the most behaviourally salient or the most physically occupied to support the maintenance of a flexible model of the environment, and this function could help explain why replays are inconsistent with a rigid recapitulation that passively records recent experience [84,87,88]. These reports suggest that replay, instead of directly reinstating specific episodes, may abstractly reflect experiences via an internal spatial map.

While there is little doubt that replays can be derived from prior experience, both in the case of a rigid recapitulation or abstract model based on the past, what remains unclear is whether neural processes within or beyond the hippocampus interpret replay events as temporally situated in the past. 

For example, a replay of recently traversed locations behind the animal, that is not subsequently traversed, is better correlated with past than future behavior, but this does not rule out the possibility that this replay represented a potential future traversal of those locations, or a spatial sequence without a projection in time. Despite this ambiguity, a replay can indeed be related to prior behavioral experiences. Moreover, these findings on replay exemplify how generative activity in the hippocampus can represent various possibilities that differ from the actual present—here, in the form of spatial paths in known environments.

In parallel to replay during rest, neural firing in the hippocampus during movement has also been suggested to be recapitulative. During movement, an internally generated network-level activity pattern, the 8 Hz theta rhythm, is observed throughout the rodent hippocampus [89–92]. Place cells are known to fire systematically about the theta rhythm, such that neurons with place fields behind, at, and ahead of the animal fire at early, intermediate, and later phases of theta cycles, respectively [55,93,94]. 

Accordingly, collective place cell firing during a single cycle can represent a series of locations consistent with sweeping from the immediate past and present ahead to anticipated future locations (rightmost example in figure 1b) [63]. Although firing in early phases of the theta rhythm can recapitulate locations just traversed by the animal, this firing appears to be consistent with the immediate actual past (for instance, as opposed to alternative past (counterfactual) locations) [63,95].
This suggests that early theta phase representations may also be best understood as reflecting experience, and not possible experience. That said, hippocampal firing during movement can correspond to locations behind the animal and is often thought to reflect the recent past [53,96,97].

(b) Representations consistent with anticipated futures

Place cell firing can also correspond to upcoming spatial paths, suggesting that generative representations may anticipate future experiences. As introduced above, place cells firing in late phases of theta cycles tend to have place fields in locations ahead of the animal [53,55]. The extent to which this activity projects ahead of the animal can correlate with the distance the animal subsequently traverses, consistent with the possibility of future anticipation or prediction [98]. When multiple paths are available (such as a path bifurcating), hippocampal firing has been found to proceed ahead along only one path at a time [68,99]. Furthermore, place cell firing corresponding to the left or right path ahead can occur on interleaved theta cycles, consistent with serially representing alternatives (figure 2a) [68]. 

These internally generated representations are consistent with generatively representing anticipated possibilities, and are reminiscent of deliberation [99]. However, while in some cases the associated neural firing can predict the animal’s subsequently taken path [99–101], firing patterns associated with alternation between paths fail to reliably predict the animal’s subsequent choice [68,99].

Apart from generative activity associated with theta, replays suggestive of anticipated future experiences have also been reported. In early work, replay was found to correspond to sequences of locations starting near and projecting ahead of the animal, just before running along that same path in the linear maze, consistent with anticipation of upcoming experience [74,79,80]. Since then, several studies have reported that replay in environments with more options (an open arena or multi-arm maze) is biased toward goal locations that the animal subsequently visits [102,103]. While replay can indeed correspond to subsequently taken paths, recent work from our group shows that replay fails to predict upcoming choices [82].

Seeking to relate generative firing to behavioral episodes in subjects’ past or future (e.g. the choice of maze arm in the previous or next trial) has been a common approach in investigating the contributions of hippocampal activity to cognitive functions, especially past-oriented functions such as episodic recall and future-oriented functions such as planning. Task paradigms that disambiguate before upcoming experience are well suited for this approach [82]. However, relating generative neural activity to particular locations behaviourally occupied in the past and future does not necessarily indicate that such activity is an internal representation that refers temporally to the past or future. 

For example, neural firing corresponding to one of two paths ahead of the subject is consistent with a possible future, yet may also reflect recall of a prior traversal of that location, or simply not have any reference in time. In this sense, it remains an open question whether generative firing patterns observed in the hippocampus can refer to experiences projected into the future. Apart from this, it remains the case that some instances of generative firing during theta and replay can correspond to potential future locations, and may thereby contribute to explicitly anticipatory functions such as planning.

For more information:1950477648nn@gmail.com