ABSTRACT
The so called Good Old Fashioned Artificial Intelligence has established itself firmly in the past decades as a research methodology. The methods and tools it uses are symbolic, highly structured representations of domain knowledge and transformations of these representations by means of formally stated rules. These rule based theories can function (and are vital) as abstract formal descriptions of aspects of cognition, constraining any cognitive theory. However, some authors go beyond that and claim that metal processes are symbolic operations performed on mental representations of rules (see Fodor, 1975). Until the connectionist paradigm emerged there was no real alternative to that view. But now, in this new paradigm, the departure from reliance on the explicit mental representation of rules is central, and thus the conception of cognition is fundamentally different (Bharucha & Olney, 1989). This holds regardless of the fact that the behavior of connectionist models could be formally described in rules. These distributed models
consistent of a large number of simple elements, or cells, each of which has its own activation level. These cells are interconnected in a network, the connections serving to excite or inhibit others. Connectionism opened the possibility of defining models which have characteristics that are hard to achieve in traditional AI, in particular robustness, flexibility and the possibility of learning (Rumelhart & McClelland, 1986). The connectionist boom brought forth much interesting work, also in the field of music (Todd & Loy, forthcoming). Although many researchers lost their critical attitude, impressed by the good performance of some (prototypical) models, it became soon clear that more study was needed to the limitations of these models. A connectionist model that ‘works” well, constitutes in itself no scientific progress, when questions like the sensitivity to parameter changes, the scalability to larger problems and the dependancy of the model on a specific input representation, cannot be answered. However, it is possible to describe the behavior of a connectionist model from different abstract perspectives that provide more insight in its limitations and its validity as a cognitive model than simulations or test runs alone. These perspectives are also fruitful for the analysis of traditional AI models. In this article we pursue this approach for a connectionist and a traditional AI model of rhythm perception as a case study for the wider issue how the paradigms themselves relate.
