Commentary on Hurford

Abstract: 58 words

Main text: 985 words

References: 193 words

Entire text: 1283 words

Predicates as cantilevers for the bridge between perception and knowledge

Gregory V. Jones

Department of Psychology

University of Warwick

Coventry CV4 7AL

U.K.

+44 24 7652 3812

G.V.Jones@warwick.ac.uk

http://www.warwick.ac.uk/fac/sci/Psychology/staff/academic.html#GJ

Abstract

The predicate-argument approach, focussed on perception, is compared with the ease-of-predication (or predicability) approach, focussed on encyclopedic knowledge. The latter offers functional prediction and implementation in connectionist models. However, the two approaches characterise predicates in different ways. They thus resemble predicational cantilevers built out from opposite sides of cognition, with a gap that is yet to be bridged.

Summarising his approach, Hurford notes in his antepenultimate sentence that "It remains to provide an explanation for the typical structure of modern languages, organized around the Noun/Verb dichotomy" (Section 6.3), and hints in his penultimate sentence that the explanation may turn out to be exemplified by the various properties of being a dog. This provides an irresistible cue to highlight an alternative predicational account which encompasses not only the theoretical dichotomy but also, strangely enough, the canine properties. This is the Ease of Predication (or Predicability) approach (e.g., Jones, 1985, 2002).

Whereas Hurford's predicate-argument approach focuses on the relation between predication and perception, the ease-of-predication approach focuses on the relation between predication and encyclopedic knowledge. The ease-of-predication approach makes detailed predictions at a behavioral level by virtue of its characterisation of differences among words in their psychological potential. Normative values of a word's predicability are established on the basis of people's judgements regarding the ease of making statements about the word's referent. The standard instructions (see de Mornay Davies & Funnell, 2000; Jones, 1985) include the following: "As an example, the word 'dog' would probably be judged as very easy to make simple factual statements about, because it can readily be put into statements such as the following:

A dog is a type of animal.

A dog often lives in a kennel.

A dog barks when angry.

A dog can be pedigree or mongrel.

A dog has four legs

A dog is called a puppy when young.

A dog wags its tail when pleased.

A dog can be as small as a chihauhau.

A dog can be as big as a St. Bernard.

A dog sometimes chases a cat."

(Jones, 1985, p. 6).

The predicates in the preceding illustration may at first sight seem sufficiently varied to make a logician blench, but the variety is intentional because the purpose is to index the ease of access to the whole range of encyclopedic knowledge, unconfined by any particular procrustean formalism. Thus in logic the examples may be viewed as encompassing predicates that are both simple (e.g., is a type of animal) and relational (e.g., can be as big as a St. Bernard), and also as ranging over the alethic modalities of both necessity (e.g., has four legs) and possibility (e.g., often lives in a kennel). Despite this diversity, when an ease-of-predication judgement for each of a set of words has been established in this way, it turns out to be predictive of several other properties of the word, including the likelihood that it is successfully read by people with deep dyslexia and, for people in general, the likelihood that it successfully retrieves information from memory and that it evokes the experience of mental imagery (e.g., Jones, 2002; Williams, Healy, & Ellis, 1999). Furthermore, the method of assessment of ease of predication is not confined to self-report judgements. Similar results are obtained when assessment is made instead in terms of the response time taken to produce such statements behaviorally (e.g., Jones, 2002; Williams at al., 1999), with a high correlation (negative in sign, of course) between words' predication times and predicability judgements.

Are the predicate-argument and ease-of-predication approaches compatible? Having been cantilevered out from opposite sides of the cognitive landscape--respectively, from perception/action and from knowledge/language--it seems reasonable to hope that these two stretches of predicational bridge can be made to join up in the middle. However, there is a problem in the shape of their differing conceptions of predication. In the case of the predicate-argument approach, the interpretation is derived from first order predicate logic. In the case of the ease-of-predication approach, the interpretation derives most naturally from the linguistic concept of predication as comment upon a topic: "The most general characterization of predicative constructions is suggested by the terms 'topic' and 'comment' for their [immediate constituents]: the speaker announces a topic and then says something about it." (Hockett, 1958, p. 201). It is rather as though one half of the bridge were designed in metres and the other in yards. What are the incentives for attempting to overcome the difference in gauge?

From the point of view of the ease-of-predication approach, the theoretical parsimony of the predicate-argument approach offers the prospect ultimately of an elegant reductionism. From the point of view of the predicate-argument approach, the most obvious attraction of the ease-of-predication approach is its compatability with the noun/verb dichotomy--a dichotomy to which Hurford noted that his present formulation appeared to pay little recognition. However, the ease-of-predication approach also offers access to a range of explanatory possibilities at both the functional and neural levels. At one level, it has been shown to lend itself naturally to implementation in the form of artificial neural networks (e.g., Hinton, Plaut, & Shallice, 1993; Plaut, 1999). Thus the influential connectionist model of deep dyslexia proposed by Plaut and Shallice (1993) relies upon semantic features distributed in accordance with the ease-of-predication approach, employing only two features for "past" ("has duration" and "refers to a previous time") but sixteen features for "post" (e.g., "found on farms" and "used for games or recreation"). At another level, the ease-of-predication approach appears to offer satisfactory functional explanations for findings which have previously been interpreted in terms of mental imagery (e.g., Paivio, 1971, 1983). If, for example, either "post" or "past" were now to be used as a retrieval cue to assist the reader in remembering the other, associated word, then previous experimental evidence would suggest that "post", the word with higher predicability--and imageability--would be more likely to succeed in retrieving "past" than vice versa, a finding that can readily be explained in terms of the greater availability of predicational routes from "post", but is difficult to account for in terms of imagery as such. It would be valuable if it were to prove possible to complete a predicational bridge between such findings and Hurford's characterisation of the elements of perception.

References

de Mornay Davies, P., & Funnell, E. (2000). Semantic representation and ease of predication. Brain and Language, 73, 92-119.

Hinton, G. E., Plaut, D. C., & Shallice, T. (1993). Simulating brain damage. Scientific American, 269(4), 58-65.

Hockett, C. F. (1958). A course in modern linguistics. New York: Macmillan.

Jones, G. V. (1985). Deep dyslexia, imageability, and ease of predication. Brain and Language, 24, 1-19.

Jones, G. V. (2002). Predicability (ease of predication) as semantic substrate of imageability in reading and retrieval. Brain and Language, 82, 159-166.

Paivio, A. (1971). Imagery and verbal processes. New York: Holt, Rinehart & Winston.

Paivio, A. (1983). The empirical case for dual coding. In J. C. Yuille (Ed.), Imagery, memory and cognition (pp. 307-332). Hillsdale, NJ: Erlbaum.

Plaut, D. C. (1999). A connectionist approach to word reading and acquired dyslexia: Extension to sequential processing. Cognitive Science, 23, 543-568.

Plaut, D. C., & Shallice, T. (1993). Deep dyslexia: A case study of connectionist neuropsychology. Cognitive Neuropsychology, 10, 377-500.

Williams, J. M. G., Healy, H. G., & Ellis, N. C. (1999). The effect of imageability and predicability of cues in autobiographical memory. Quarterly Journal of Experimental Psychology, 52A, 555-579.