Hierarchical cognitive maps

Exploring the Structure of Spatial Representations

It has been suggested that the map-like representations that support human spatial memory are fragmented into sub-maps with local reference frames, rather than being unitary and global. However, the principles underlying the structure of these ‘cognitive maps’ are not well understood. We propose that the structure of the representations of navigation space arises from clustering within individual psychological spaces, i.e. from a process that groups together objects that are close in these spaces. Building on the ideas of representational geometry and similarity-based representations in cognitive science, we formulate methods for learning dissimilarity functions (metrics) characterizing participants’ psychological spaces. We show that these learned metrics, together with a probabilistic model of clustering based on the Bayesian cognition paradigm, allow prediction of participants’ cognitive map structures in advance. Apart from insights into spatial representation learning in human cognition, these methods could facilitate novel computational tools capable of using human-like spatial concepts. We also compare several features influencing spatial memory structure, including spatial distance, visual similarity and functional similarity, and report strong correlations between these dimensions and the grouping probability in participants’ spatial representations, providing further support for clustering in spatial memory.

Computational cognitive models of spatial memory in navigation space: a review

Spatial memory refers to the part of the memory system that encodes, stores, recognizes and recalls spatial information about the environment and the agent's orientation within it. Such information is required to be able to navigate to goal locations, and is vitally important for any embodied agent, or model thereof, for reaching goals in a spatially extended environment. In this paper, a number of computationally implemented cognitive models of spatial memory are reviewed and compared. Three categories of models are considered: symbolic models, neural network models, and models that are part of a systems-level cognitive architecture. Representative models from each category are described and compared in a number of dimensions along which simulation models can differ (level of modeling, types of representation, structural accuracy, generality and abstraction, environment complexity), including their possible mapping to the underlying neural substrate. Neural mappings are rarely explicated in the context of behaviorally validated models, but they could be useful to cognitive modeling research by providing a new approach for investigating a model's plausibility. Finally, suggested experimental neuroscience methods are described for verifying the biological plausibility of computational cognitive models of spatial memory, and open questions for the field of spatial memory modeling are outlined.

Spatial learning and action planning in a prefrontal cortical network model

The interplay between hippocampus and prefrontal cortex (PFC) is fundamental to spatial cognition. Complementing hippocampal place coding, prefrontal representations provide more abstract and hierarchically organized memories suitable for decision making. We model a prefrontal network mediating distributed information processing for spatial learning and action planning. Specific connectivity and synaptic adaptation principles shape the recurrent dynamics of the network arranged in cortical minicolumns. We show how the PFC columnar organization is suitable for learning sparse topological-metrical representations from redundant hippocampal inputs. The recurrent nature of the network supports multilevel spatial processing, allowing structural features of the environment to be encoded. An activation diffusion mechanism spreads the neural activity through the column population leading to trajectory planning. The model provides a functional framework for interpreting the activity of PFC neurons recorded during navigation tasks. We illustrate the link from single unit activity to behavioral responses. The results suggest plausible neural mechanisms subserving the cognitive “insight” capability originally attributed to rodents by Tolman & Honzik. Our time course analysis of neural responses shows how the interaction between hippocampus and PFC can yield the encoding of manifold information pertinent to spatial planning, including prospective coding and distance-to-goal correlates.

We study spatial cognition, a high-level brain function based upon the ability to elaborate mental representations of the environment supporting goal-oriented navigation. Spatial cognition involves parallel information processing across a distributed network of interrelated brain regions. Depending on the complexity of the spatial navigation task, different neural circuits may be primarily involved, corresponding to different behavioral strategies. Navigation planning, one of the most flexible strategies, is based on the ability to prospectively evaluate alternative sequences of actions in order to infer optimal trajectories to a goal. The hippocampal formation and the prefrontal cortex are two neural substrates likely involved in navigation planning. We adopt a computational modeling approach to show how the interactions between these two brain areas may lead to learning of topological representations suitable to mediate action planning. Our model suggests plausible neural mechanisms subserving the cognitive spatial capabilities attributed to rodents. We provide a functional framework for interpreting the activity of prefrontal and hippocampal neurons recorded during navigation tasks. Akin to integrative neuroscience approaches, we illustrate the link from single unit activity to behavioral responses while solving spatial learning tasks.

Multihierarchical interactive task planning: application to mobile robotics

To date, no solution has been proposed to human-machine interactive task planning that deals simultaneously with two important issues: 1) the capability of processing large amounts of information in planning (as it is needed in any real application) and 2) being efficient in human-machine communication (a proper set of symbols for human-machine interaction may not be suitable for efficient automatic planning and vice versa). In this paper, we formalize a symbolic model of the environment to solve these issues in a natural form through a human-inspired mechanism that structures knowledge in multiple hierarchies. Planning with a hierarchical model may be efficient even in cases where the lack of hierarchical information would make it intractable. However, in addition, our multihierarchical model is able to use the symbols that are most familiar to each human user for interaction, thus achieving efficiency in human-machine communication without compromising the task-planning performance. We formalize here a general interactive task-planning process which is then particularized to be applied to a mobile robotic application. The suitability of our approach has been demonstrated with examples and experiments.

Thoughts, behaviour, and brain dynamics during navigation in the real world

How does the human brain allow us to interact with and navigate through a constantly changing world? Whilst controlled experiments using functional brain imaging can give insightful snapshots of neuronal responses to relatively simplified stimuli, they cannot hope to mirror the challenges faced by the brain in the real world. However, trying to study the brain mechanisms supporting daily living represents a huge challenge. By combining functional neuroimaging, an accurate interactive virtual simulation of a bustling central London (UK), and a novel means of 'reading' participants' thoughts whilst they moved around the city, we ascertained the online neural correlates underpinning navigation in this real-world context. A complex choreography of neural dynamics was revealed comprising focal and distributed, transient and sustained brain activity. Our results provide new insights into the specific roles of individual brain areas, in particular the hippocampus, retrosplenial, and frontal cortices, as well as offering clues about how functional specialisations operate within dynamic brain systems.

Online formation of a hierarchical cognitive map for object-place association by theta phase coding

Object-place associative memory, the storage of object and place conjunctions based on a one-time experience, is hippocampal-dependent in humans. Theta phase precession, a type of neural dynamics observed in the rat hippocampus, has recently been suggested to serve a role in instantaneous memory formation based on a one-time experience, while its functional role in associating distinct types of information (object and place information) is unclear. In this study, we hypothesize that theta phase encoding with theta phase precession contributes to the storage of object-place associations. To examine this hypothesis, we propose a neural network model of the corticohippocampal system, including central-peripheral visual pathways and theta phase coding in the hippocampus. Memory storage computer experiments demonstrate that the hippocampal network successfully stores the object-place associations of a one-time experience. Interestingly, it is also found that a random visual input sequence results in a robust formation of asymmetric connections between objects and scenes instantaneously after a single trial. Furthermore, it is found that scene-object connections and scene-scene connections form a hierarchical network representing the spatial alignment of scenes and objects in the environment. Our findings indicate that the theta phase coding, as observed in the rat hippocampus, can facilitate the online memory storage of complex environments in humans as a hierarchical cognitive map.

Modeling mental navigation in scenes with multiple objects

Various lines of evidence indicate that animals process spatial information regarding object locations differently from spatial information regarding environmental boundaries or landmarks. Following Wang and Spelke's (2002) observation that spatial updating of egocentric representations appears to lie at the heart of many navigational tasks in many species, including humans, we postulate a neural circuit that can support this computation in parietal cortex, assuming that egocentric representations of multiple objects can be maintained in prefrontal cortex in spatial working memory (not simulated here). Our method is a generalization of an earlier model by Droulez and Berthoz (1991), with extensions to support observer rotation. We can thereby simulate perspective transformation of working memory representations of object coordinates based on an egomotion signal presumed to be generated via mental navigation. This biologically plausible transformation would allow a subject to recall the locations of previously viewed objects from novel viewpoints reached via imagined, discontinuous, or disoriented displacement. Finally, we discuss how this model can account for a wide range of experimental findings regarding memory for object locations, and we present several predictions made by the model.