Exploration, Navigation and Cognitive Mapping

Voicu, Horatiu; Schmajuk, Néstor A.

doi:10.1177/105971230000800301

Cited by 15 publications

(9 citation statements)

References 31 publications

(38 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Some key elements of the endotaxis model have appeared in prior work, starting with the notion of ascending a scalar goal signal during navigation [50, 52, 66]. Several models assume the existence of a map layer, in which individual neurons stand for specific places, and the excitatory synapses between neurons represent the connections between those places [23, 33, 39, 47, 53, 65, 66]. Then the agent somehow reads out those connections in order to find the shortest path between its current location (the start node) and a desired target (the end node).…”

Section: Discussionmentioning

confidence: 99%

“…The most popular scheme is to somehow inject a signal into the desired end node, let it propagate backward through the network, and read out the magnitude or gradient of the signal near the start node [23, 25, 26, 33, 39, 47]. In general this requires some accessory system that can look up which neuron in the map corresponds to the desired end node, and which neuron to the agent’s current location or its neighbors; often these accessory functions remain unspecified [33, 53, 66]. By contrast, in the endotaxis model the signal is propagated in the forward direction starting with the activity of the place cell at the agent’s current location.…”

Section: Discussionmentioning

confidence: 99%

See 1 more Smart Citation

Endotaxis: A neuromorphic algorithm for mapping, goal-learning, navigation, and patrolling

Zhang

Rosenberg

Perona

et al. 2021

Preprint

View full text Add to dashboard Cite

An animal entering a new environment typically faces three challenges: explore the space for resources, memorize their locations, and navigate towards those targets as needed. Experimental work on exploration, mapping, and navigation has mostly focused on simple environments - such as an open arena, a pond [1], or a desert [2] - and much has been learned about neural signals in diverse brain areas under these conditions [3,4]. However, many natural environments are highly constrained, such as a system of burrows, or of paths through the underbrush. More generally, many cognitive tasks are equally constrained, allowing only a small set of actions at any given stage in the process. Here we propose an algorithm that learns the structure of an arbitrary environment, discovers useful targets during exploration, and navigates back to those targets by the shortest path. It makes use of a behavioral module common to all motile animals, namely the ability to follow an odor to its source [5]. We show how the brain can learn to generate internal "virtual odors'" that guide the animal to any location of interest. This endotaxis algorithm can be implemented with a simple 3-layer neural circuit using only biologically realistic structures and learning rules. Several neural components of this scheme are found in brains from insects to humans. Nature may have evolved a general mechanism for search and navigation on the ancient backbone of chemotaxis.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Discussionmentioning

confidence: 99%

Endotaxis: A neuromorphic algorithm for mapping, goal-learning, navigation, and patrolling

Zhang

Rosenberg

Perona

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…• A model also based on self-organized learning was proposed by Voicu (2003), extending their earlier work (Voicu & Schmajuk, 2000). Unlike the model above, it is capable of running in a full two-dimensional metric simulation instead of a restricted mazelike environment.…”

Section: Models Evaluated In Simulationsmentioning

confidence: 97%

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

et al. 2015

View full text Add to dashboard Cite

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.

show abstract

“…In order to demonstrate the interaction of the locometric and cognitive map spaces, we tested the model in a linear maze with four nodes and food in the farthest node (Figure 12, middle). (The model bears some similarity to that of Voicu & Schmajuk, 2000, but they fail to distinguish the locometric map from the more abstract map of the kind provided by WG.) The model was trained for 20 trials on the linear maze, which contained food in one end.…”

Section: Spatial Difference Learning and The Dual (At Least) Charactementioning

confidence: 98%

Multiple levels of spatial organization: World Graphs and spatial difference learning

Arbib

Bonaiuto

2012

Adaptive Behavior

View full text Add to dashboard Cite

It is often suggested that the place cells of the hippocampus (and more recently, the grid cells of the entorhinal cortex) furnish a cognitive map. However, this can only be part of the story: (1) the ''hippocampal chart'' provided by place cells and grid cells differs radically when a rat is placed in different environments and so a higher level organization is needed to link these charts into an overall cognitive map of the rat's world; and (2) even without a hippocampus a rat can exploit much of the spatial structure of its world. The World Graph (WG) model addressed the former problem, whereas the Taxon Affordance Model (TAM) was developed to address the latter, with the two models being integrated to form the TAM-WG model. Here we provide a new version strengthened in three ways: (1) we relate the TAM model to explicit ideas of executability and desirability; (2) we show how temporal difference learning elegantly supplies ''spatial difference learning'' to resolve the debate between the local hypothesis and the non-local hypothesis for node selection in the original WG model; and (3) we analyze an explicit example of how the ''locometric map'' provided by grid cells and place cells can complement the high-level cognitive map given by the WG, demonstrating the importance of navigation algorithms that integrate across multiple levels of spatial organization.

show abstract

Exploration, Navigation and Cognitive Mapping

Cited by 15 publications

References 31 publications

Endotaxis: A neuromorphic algorithm for mapping, goal-learning, navigation, and patrolling

Endotaxis: A neuromorphic algorithm for mapping, goal-learning, navigation, and patrolling

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

Multiple levels of spatial organization: World Graphs and spatial difference learning

Contact Info

Product

Resources

About