Minds and Computers : An Introduction to the Philosophy of Artificial Intelligence
partly (although not entirely) informed by beliefs concerning the
Download 1.05 Mb. Pdf ko'rish
|
document (2)
partly (although not entirely) informed by beliefs concerning the nature of mental representation. 18.1 INTENTIONALITY Intentionality is the technical philosophical term for the representa- tional nature of mental states. Intentional states are those which are about something, which represent something. 181 The terms ‘intentionality’ and ‘intentional’ are not to be confused with the verb ‘intend’ and its cognates. Whether or not a state is inten- tional, in the technical philosophical sense, has nothing to do with it being intended by some agent. Rather, a mental state is intentional just in case it is about something. Intentionality is the property of mental states such that they are directed towards an object of repre- sentation (a thing which is represented). It is mental representations which are the primitive bearers of intentionality. Our mental states are intentional by virtue of having mental representations as constituents. So, for instance, my belief that ‘my dog is a fine companion’ involves, inter alia, a token mental rep- resentation of ‘my dog’. My belief is about my dog by virtue of having a constituent mental representation which is about my dog. My mental representation of ‘my dog’ is directed towards (about) its object of representation – namely, my dog. This brings us to one of the important features of mental repre- sentations I want to highlight here. Mental representations are cat- egorial. My mental representation of ‘dog’ picks out all and only dogs. In other words, it serves to categorise – in my mental life – those things which I take to be dogs and distinguish them from those things which I take not to be dogs. Similarly, my mental representation of ‘brown’ picks out all and only the things I take to be brown and my mental representation of ‘chair’ picks out all and only the things I take to be chairs. The other important feature of mental representations I want to highlight is that they are compositional. Mental representations compose into more complex mental representations. Given, for instance, my possession of mental representations of ‘brown’ and ‘dog’, I need nothing further to compose the more complex mental representation of ‘brown dog’. This more complex mental represen- tation picks out all and only the things I take to be brown dogs. This compositionality of mental representations allows for one part of an account of how it is that mental representations are conferred with their intentional content. Complex mental representations inherit their intentionality from the primitive mental representations of which they are composed. The crucial – and most di fficult – question to answer, however, is how it is that primitive mental representations are conferred with their intentional content. How is it that our atomic mental representations come to be about their objects of representation? In other words, what is the nature of the relation between mental representations and the (categories of) objects they represent? 182 18.2 CATEGORIES AND CONTENT To give an account of the relation between mental representations and their intentional objects is to give part of a semantics for mental representation. There are numerous theories of the semantics of mental representation but I’m not going to give a balanced exposition of the available theories here. Instead, I want to give just the barest sketch of two kinds of theories. On one hand we have theories according to which mental represen- tations are essentially discrete. On the other hand we have theories according to which mental representations are essentially interre- lated. Theories of this first kind fit well with the symbolic paradigm in artificial intelligence research. According to this kind of theory, mental representations are symbols. Theories of the second kind fit well with the connectionist para- digm in artificial intelligence research. According to this kind of theory, mental representations are distributed patterns. A commitment to symbolic representation, on the one hand, or distributed representation on the other, brings with it a raft of corol- lary commitments. These include commitments concerning the mech- anisms by which representations are conferred with their intentional content, the nature and structure of the categories represented and the ways in which mental representations interact. Proponents of symbolic representation take representations to be essentially discrete in a number of ways. The mechanism by which symbols are conferred with their content is understood as some kind of direct relation between tokens of the symbol and objects of repre- sentation. Crucially, this mechanism is such that the content of a symbol in no way depends on the content of other symbols. Each symbol is discretely conferred with its intentional content. Furthermore, symbolic representations are understood to remain discrete in their interactions with other representations. The compos- itionality of mental representation is understood to be simple syn- tactic concatenation. When symbols compose to give more complex representations, each symbol always brings the same content to the complex in which it participates. In other words, the content of symbols is taken to be contextually insensitive. It is a further feature of symbols that their presence is binary – a symbol token is either present or not. If a symbol token is present, it is fully present and if it is absent, it is fully absent. Symbols, if you will, are either on or o ff, with no scope for anything in between. This 183 binary nature of symbolic representation has implications pertaining to the nature of the categories which they represent. If mental representations are symbolic then the categories which they represent must admit of sharp borders and no internal structure. In other words, if mental representations are symbolic, then the cat- egories they represent are like boxes – objects are either in the cat- egory or not and there are no better or worse cases of category membership. Proponents of symbolic representation, to recap, take the content conferring mechanism on representations to be discrete, the cat- egories they represent to be binary and unstructured, and the com- position of mental representation to be contextually insensitive syntactic concatenation. Advocates of distributed representation, on the other hand, have a di fferent understanding of each of these elements of the semantics of mental representation. 18.3 SYMBOLS AND PATTERNS Theorists who endorse an account of mental representations as dis- tributed patterns understand representations to be essentially interre- lated. The semantics of mental representation that such a theorist will advance are such that the mechanism by which content is conferred on a representation is essentially mediated by relations to other rep- resentations. There are a number of ways to flesh out this mediation but the mechanics of particular theories need not concern us here. What is important for our purposes is the commitment to the interrelated nature of mental representations, in stark opposition to the view held by proponents of symbolic representation. The way in which a mental representation is conferred with its intentional content, according to distributed accounts of representation, is essentially bound up with the way in which other mental representations are conferred with their intentional content. The composition of distributed mental representations is also taken to be somewhat more complex than syntactic concatenation. It is crucial that there be an account of the composition of mental rep- resentation that is su fficient to secure the systematicity of cognition, since this is generally held to account for the productivity of the lin- guistic facility. Those who endorse a view of mental representation as distributed understand the composition of representations to be the highly 184 complex interaction of patterns of activation in a distributed network. Precisely what this means will become clearer in the follow- ing chapter when we discuss artificial neural networks. For present purposes, it su ffices to appreciate that the way distributed representa- tions compose is contextually modulated. In other words, the content that a particular representation brings to the complex representation in which it participates will vary in a way that is dependent on the other particular representations also participating in the complex. It is a further feature of distributed representations that they can be partially tokened. Since token representations are taken to be pat- terns of activation widely distributed across an interconnected network of nodes, these patterns can be partially activated. Again, precisely what this means will become clearer in the following chapter. The important point here is an appreciation of the implications that the possibility of partial tokening of mental representations has with respect to the nature of the categories represented. If mental representations are distributed patterns which can be Download 1.05 Mb. Do'stlaringiz bilan baham: |
Ma'lumotlar bazasi mualliflik huquqi bilan himoyalangan ©fayllar.org 2024
ma'muriyatiga murojaat qiling
ma'muriyatiga murojaat qiling