Commentary on: “The neural basis of predicate argument structure” by Jim Hurford
Word counts:
Abstract: 68
Text: 1082
References: 84
Total: 1277
Cognitive structure, logic, and language
Grant Gillett
Otago Bioethics Centre
64-3-4740999
grant.gillett@stonebow.otago.ac.nz
Abstract:
Philosophical accounts of thought crucially involve an array of abilities to identify general properties or features of the world (corresponding to concepts) and objects which instance those general properties. Abilities of both types can be grounded in a naturalistic account of the usefulness of cognitive structures in adaptive behaviour. Language enhances these abilities by multiplying the experience bases giving rise to them and helping to overcome subjective biases.
Main
Text:
Hurford’s paper signals a potentially important resonance between Fregean truth conditional semantics and neural structure of the type that philosophers like Evans (1982) and Campbell (1995) have begun to explore.
Underpinning this post-Fregean stream of philosophical naturalism is the idea that thought works with a world-picture comprising objects and their properties and that such a cognitive map of the world is an adaptive achievement for any higher animal. According to this view the crucial adaptation is a move away from feature driven stimulus-response patterns to tracking abilities based on more complex response sets related to object characterisation and recognition. It is a further move for any creature to share information from multiple subject-object encounters so that a rich dossier of information can be used to inform its behaviour in relation to the world in which it lives. A theory of thought and its content of this kind is indicated by the naturalistic tendency of Wittgenstein’s later (post-Tractatus) writings and allows truly innovative thinking in theoretical cognitive neuroscience.
We can begin with truth conditional semantics according to which the semantic content of a thought or sentence is given by the conditions in which a competent thinker would hold it to be true. For a creature in a world like ours such conditions would, stereotypically and ideally, concern an object and a predicate: <the cat is moving>; <the poison berry is red>; and so on. Notice that even if the noun phrase in the sentence concerned is complex (e.g. “the poison berry”) it concerns an object that is salient for the creature thinking the thought. The simplest question that any creature could ask the world would be <what is that> where an object catches the creature’s attention and it must assign properties to the object so as to potentiate a range of appropriate responses. And the information available to the creature is greatly enhanced if objects can be identified and reidentified (under conditions which yield slightly different perceptual features from the first encounter) so that a dossier for objects of the relevant type can be compiled.
It would be a further step to identify numerical individuals (Blackie) rather than categorical types (big monkey) and the computational resources for such a task would be useful wherever the identification and reidentification of individuals had an adaptive pay-off (for instance where appropriate responses differed individuals of the same type as in a dominance hierarchy). Human beings could be expected to have an individual identification system good enough to meet the needs of the environment in which their language developed because exactly this condition holds. These thoughts suggest a three level scheme of the type mooted in Hurford’s article but divorced from the implicit Cartesianism of Frege and many empiricists.
Stage 1: egocentric space and stimulus response pairings based in individual dispositions – a feature–response scheme.
Stage 2: egocentric space and activity directed by information about objects as loci of salient general properties -- the simple PREDICATE(x) scheme.
Stage 3: activity (cognitive or overt) in a shared world directed on mutually accessible identified objects and open to information from others – PREDICATE(x’)
Human beings have language and
interpersonal behaviour as a primary mode of adaptation (Trevarthen,
2001) so that stages 2 and 3 of Hurford’s scheme are
modified by signs and conventional referring devices for both general features of the world and particular objects in the
world. These facilitate and elaborate the response repertoire of any individual
so as to give them access to shared dossiers of information about objects and
their affordances. If language is used to expand the
informational power of an organism, we need to supplement the 3 stage scheme by
noting the effect of semantic markers (Gillett & McMillan, 2000). True
concepts and mature conceptions of objects are tied to truth conditions by the
normative uses of natural language so that there is a concurrence of semantic
content between co-linguistic speakers. Thus, early in language learning I
might think that a dog is a big black furry thing that bounds around the
neighbourhood but later I discover that dogs include chihuahuas
and poodles. Such convergence in categorisation with other competent language
users occurs by conversational correction within a co-linguistic human group. By
noticing this fact, we can, without denying the continuity between human
thought and that of higher animals, bring out a point of difference which
increases the power of human epistemic activity and in which language plays a
central role.
Attention, a prominent theme in Luria’s work (1973), is important in the formation of concepts and conceptions of objects. In both cases the subject must ground the thought concerned on selected aspects of the environment (Gillett, 1992). For instance, in the PREDICATE(x) type thought <that frog is bright orange>), <that frog> focuses on and tracks an object, and <bright orange> links a feature of the frog to other stimulus arrays instancing that colour. Kant said that thinking was “cognition through concepts”(B94) whereby information from an object was linked in two distinct ways to form discursive or semantic content: (i) to general concepts (square, red, dog, baroque) and (ii) to other presentations of the same object. The second link could, in the animal case, be mediated by biological abilities to track the object concerned but in humans this would be vastly enhanced by the language-related practice of naming things or using denoting expressions (such as definite descriptions – “the black cat from next door”). These, as Hurford notes are fundamentally deictic or indexical and need only be specific enough to work in the context where they are used (thus they may not have the logical qualities of unique designators). Wittgenstein’s version of naturalism implies that such indexical expressions are “as good as it gets” for embodied creatures banging about in a finite domain.
The present account implies that logic is an idealization of natural language which is a tool to aid our cognitive activity which picks out objects and designates their significant properties. As such logic formalizes our thought content as it appears in language (transformed by the conventional devices that are intrinsic to natural languages). Therefore logic embodies the fundamental difference between two types of element that make up our cognized world – objects and the properties they instance. It is worth emphasising that a cognitive map of our environment which can represent relatively enduring objects and their properties (including typical trajectories through time and space) is likely to be of great value as we try to make use of the opportunities presented by our world.
References:
Campbell, J. (1994) Past, space, and self.
Evans G.
(1982) Varieties of reference
Gibson, J.
(I979). The Ecological
Approach to Visual Perception.
Gillett, G.
(1992) Representation Meaning and Thought.
Kant,
Luria, A.R., (1973) The working brain Harmondsworth; Penguin
Trevarthen, c. (2001) The
neurobiology of early communication: Intersubjective
regulations in human brain development Handbook of brain and behaviour in human development : 841-881.