We propose a novel application of Formal Concept Analysis (FCA) to neural decoding: instead of just trying to figure out which stimulus was presented, we demonstrate how to explore the semantic relationships in the neural representation of large sets of stimuli. FCA provides a way of displaying and interpreting such relationships via concept lattices. We explore the effects of neural code sparsity on the lattice. We then analyze neurophysiological data from high-level visual cortical area STSa, using an exact Bayesian approach to construct the formal context needed by FCA. Prominent features of the resulting concept lattices are discussed, including hierarchical face representation and indications for a product-of-experts code in real neurons. 1

The inferior temporal (IT) cortex in monkeys plays a central role in visual object recognition and learning. Previous studies have observed patches in IT cortex with strong selectivity for highly familiar object classes (e.g., faces), but the principles behind this functional organization are largely unknown due to the many properties that distinguish different object classes. To unconfound shape from meaning and memory, we scanned monkeys with functional magnetic resonance imaging while they viewed classes of initially novel objects. Our data revealed a topography of selectivity for these novel object classes across IT cortex. We found that this selectivity topography was highly reproducible and remarkably stable across a 3-month interval during which monkeys were extensively trained to discriminate among exemplars within one of the object classes. Furthermore, this selectivity topography was largely unaffected by changes in behavioral task and object retinal position, both of which preserve shape. In contrast, it was strongly influenced by changes in object shape. The topography was partially related to, but not explained by, the previously described pattern of face selectivity. Together, these results suggest that IT cortex contains a large-scale map of shape that is largely independent of meaning, familiarity, and behavioral task.

The unsupervised categorization of sensory stimuli is typically attributed to feedforward processing in a hierarchy of cortical areas. This purely sensory-driven view of cortical processing, however, ignores any internal modulation, e.g., by top-down attentional signals or neuromodulator release. To isolate the role of internal signaling on category formation, we consider an unbroken continuum of stimuli without intrinsic category boundaries. We show that a competitive network, shaped by recurrent inhibition and endowed with Hebbian and homeostatic synaptic plasticity, can enforce stimulus categorization. The degree of competition is internally controlled by the neuronal gain and the strength of inhibition. Strong competition leads to the formation of many attracting network states, each being evoked by a distinct subset of stimuli and representing a category. Weak competition allows more neurons to be co-active, resulting in fewer but larger categories. We conclude that the granularity of cortical category formation, i.e., the number and size of emerging categories, is not simply determined by the richness of the stimulus environment, but rather by some global internal signal modulating the network dynamics. The model also explains the salient non-additivity of visual object representation observed in the monkey inferotemporal (IT) cortex. Furthermore, it offers an explanation of a previously observed, demand-dependent modulation of IT activity on a stimulus categorization task and of categorization-related cognitive defi cits in schizophrenic patients.

This paper deals with the analytical study of coding a discrete set of categories by a large assembly of neurons. We consider population coding schemes, which can also be seen as instances of exemplar models proposed in the literature to account for phenomena in the psychophysics of categorization. We quantify the coding efficiency by the mutual information between the set of categories and the neural code, and we characterize the properties of the most efficient codes, considering different regimes corresponding essentially to different signal-to-noise ratio. One main outcome is to find that, in a high signal-to-noise ratio limit, the Fisher information at the population level should be the greatest between categories, which is achieved by having many cells with the stimulus-discriminating parts (steepest slope) of their tuning curves placed in the transition regions between categories in stimulus space. We show that these properties are in good agreement with both psychophysical data and with the neurophysiology of the inferotemporal cortex in the monkey, a cortex area known to be specifically involved in classification tasks.

According to Bowers (2009), the finding that there are neurons with highly selective responses to familiar stimuli supports theories positing localist representations over approaches positing distributed representations (including PDP models). However, his conclusions derive from an overly narrow view of the range of possible distributed representations and of the role that PDP models can play in exploring their properties. While it is true that current distributed theories face challenges in accounting for both neural and behavioral data, the proposed localist account―to the extent that it is articulated at all―runs into more fundamental difficulties. Central to these difficulties is the problem of specifying the set of entities a localist unit represents.

A key to overcoming the limitations of classical artificial intelligence and to deal well with enormous amounts of information might be brain-like computing in which distributed representations of information are processed by dynamical systems without using symbols. We present a method for such computing. We constructed an inference system using a nonmonotone neural network, which is a kind of recurrent neural network with continuous-time dynamics. This system deduces a conclusion according to state transitions of the network in which knowledge is embedded as trajectory attractors. It has the powerful ability of analogical reasoning without special treatment for exceptional knowledge. We also propose a method of linking different neurodynamical systems and show that two mutually interacting systems can process complex spatiotemporal patterns.

the date of receipt and acceptance should be inserted later Abstract This paper proposes a novel application of Formal Concept Analysis (FCA) to neural decoding: the semantic relationships between the neural representations of large sets of stimuli are explored using concept lattices. In particular, the effects of neural code sparsity are modelled using the lattices. An exact Bayesian approach is employed to construct the formal context needed by FCA. This method is explained using an example of neurophysiological data from the high-level visual cortical area STSa. Prominent features of the resulting concept lattices are discussed, including indications for hierarchical face representation and a product-of-experts code in real neurons. The robustness of these features is illustrated by studying the effects of scaling the attributes.

You might find this additional information useful... This article cites 54 articles, 30 of which you can access free at:

We perceive real-world objects as three-dimensional (3D), yet it is unknown which brain area underlies our ability to perceive objects in this way. The macaque inferotemporal (IT) cortex contains neurons that respond selectively to 3D structures defined by binocular disparity. To examine the causal role of IT in the categorization of 3D structures, we electrically stimulated clusters of IT neurons with a similar 3D-structure preference while monkeys performed a 3D-structure categorization task. Microstimulation of 3D-structure-selective IT clusters caused monkeys to choose the preferred structure of the 3Dstructure-selective neurons considerably more often. Microstimulation in IT also accelerated the monkeys ’ choice for the preferred structure, while delaying choices corresponding to the nonpreferred structure of a given site. These findings reveal that 3D-structure-selective neurons in IT contribute to the categorization of 3D objects.