Implicit Versus Explicit Representation and Intra‐ Versus Inter‐Modular Processing

Frawley raises a very interesting issue: Can we draw on lessons learnt in computer science about different types of control processes, and the distinction between'control'and'logic''(Kowalski 1979), to shed light on the human mind? Frawley concentrates discussion on unit-level control, i. e., the control of the flow of information across modules, and the representations (logic) that exist within modules. He classifies a group of developmental syndromes (e. g., Williams, autism) as suffering from an across module control problem in contrast to, e. g., SLI (specific language impaired) children, who suffer from a within module impairment of linguistic knowledge. Frawley identifies one of the requirements of across module communication as being explicit, and finds our (Dienes and Perner 1999) attempt to explicate the meaning of the implicit explicit distinction particularly helpful. Frawley gives an admirably clear and succinct rendition of our account of predication and factuality explicitness, and uses it for his purposes. Frawley suggests that intermodular processing requires predication and factuality implicitness, whereas intramodular processing can do without:''Representations that are within domains are preferentially implicit and procedural because they must apply generally (i. e., no explicit factivity).... By contrast, representations reported out of a domain have to be maximally explicit in order to be checked and used.''We wish to elaborate on this theme. Following our original article, we can illustrate predication implicitness-explicitness with the example of individual animals being classified as cats or dogs. If all that is generated is a label'cat'then although this represents the presented individual as a cat, it only makes the property of being a cat explicit and leaves it implicit that there is an individual of which this property of cat-ness is being predicated (predication implicit knowledge). On the other hand, the representation'This is a cat,''would make this predication explicit. A representation leaves its factuality implicit if it is simply taken as true by the system; it makes its factuality explicit if it can be represented as true or false or possibly true. Predication explicitness is necessary for tracking the same individual across different contexts. How might this be related to representations used within as opposed to between modules? Modules are defined to be'informationally encapsulated''(Fodor 1983); they operate only on certain types of information, they perform very specific operations on that information, and they do so online. By contrast, non-modular processing central processing, the global workspace (Baars 1988), is inferentially promiscuous, dealing in principle with any type of content domain, making that information available to in principle any type of relevant processing, with in principle information coming in from any of a number of different modules about possibly different individuals. Given this idealised description, it would seem that information communicated by one module to other modules