Cortex and Mind Chapter 1
The goal of cognitive neuroscience
According to Fuster, cognitive neuroscience seeks to discover an isomorphic relation between a spatial and temporal order in the brain, particularly the cerebral cortex, and a spatial and temporal cognitive order.
This means that we seek to identify a set of entities and processes in the cortex that have exactly the same relations to one another as a corresponding set of cognitive entities and processes.
From what is currently known about the brain and cognition, the most plausible neural and cognitive models each have a network structure.
One way to view cognitive neuroscience is as the search for correspondences (correlated relationships) between cognitive and cortical networks.
This search leads to the study of cognitive networks.
The theory of cognitive networks
Arguments for the modular paradigm come primarily from the localizationist view of cognitive function in neuropsychology and evidence for localization of low-level sensory and motor functions from neurophysiology and neuroanatomy.
Arguments for modular paradigm: 1) the cortex is divided into many areas (modules) 2) each area has a unique microscopic structure and set of connections 3) some sensory and motor areas are known to have specialized neurophysiological functions 4) disorders of cognitive function result from localized cortical lesions
Arguments for the network paradigm come from both the distributed view of cognitive function in neuropsychology and the PDP tradition (connectionism) in artificial intelligence.
Arguments for network paradigm: 1) evidence that the specialized sensory and motor cortical areas represent the lower stages of hierarchies toward cognition
2) evidence that cognitive functions depend on widely distributed cortical regions that include the specialized sensory and motor areas, but also extend well beyond them 3) the Gestalt psychology concept that visual perception emerges out of relations among sensory elements 4) Lashley’s conclusions that memory is distributed throughout the cortex and that the neuronal substrate of memory serves many different cognitive functions 5) Hebb’s idea that distributed knowledge representation depends on cortical networks of neuronal assemblies 6) Friedrich Hayek’s (1952) proposal that percepts and memories are represented in large-scale cortical networks 7) Gerald Edelman & Vernon Mountcastle’s (1978) theory that learning, memory, and perception are widely distributed in interconnected cortical modules 8) Evidence from the field of cybernetics that cortical networks have properties of feedback systems 9) Evidence that cortical networks have self-organizing associative memory properties, such as studied by Teuvo Kohonen
The neuroscience of cognitive networks
3 fields of empirical neuroscience contribute to our understanding of cognitive networks: 1) Cortical axonal connectivity: the connectivity between cortical areas is essential to models of cortical network cognition 2) Cortical electrophysiology: a. macroscopic recording reveals timing of, and cortical regions of, activity associated with cognitive functions i) EEG & ERP recorded with macroscopic electrodes ii) MEG & ERF recorded with SQUID sensors b. microscopic recording (single-unit, multi-unit, local field potential) reveals timing of, and cellular mechanisms associated with, cognitive functions 3) Cortical neuroimaging: reveals the spatial distribution of brain activity related to cognitive functions with good spatial resolution. Problems with interpretation: the BOLD (Blood Oxygen Level-Dependent ) signal measures tissue blood perfusion; the exact relation of blood perfusion to neural activity is not known and the time resolution of this hemodynamic response is poor
The Cognit
This term is a neologism, meaning it was created by the author. It is defined as: any representation of knowledge in the brain, the cerebral cortex in particular.
Why do we need a new term for the representation of knowledge in the brain?
From its definition, it is clear that the cognit refers to a neural structure. However, if it represents knowledge, then it must be isomorphic with a cognitive structure. The creation of the new term “cognit” derives specifically from this postulated isomorphic relation, which has not previously been made explicit.
Properties of the cognit: 1) All cognitive functions consist in transactions within and between cognits 2) A cognit has a cortical network structure made up of elementary sensory and motor representations 3) These smaller units of representation constitute the nodes of the network; the nodes have their own internal network structure 4) A cognit is defined by its component nodes and the relations between them; some elementary cognits may be innate 5) Each cognit has a specific cortical network structure (topographic specificity), but cognitive functions do not – the cognitive function operates on a set of active cognits 6) Cognits are dynamic structures that change with experience 7) Learning involves the creation of new cognits, a process that includes recomposition and decomposition of existing cognits