From apples and oranges to symbolic dynamics: a framework for conciliating notions of cognitive representation (original) (raw)
Related papers
Cognitive Science, Representations and Dynamical Systems Theory
Studies of Nonlinear Phenomena in Life Science, 2003
In this paper we point out that the assumption of representation in the explanations and models of cognitive science has several disadvantages. We propose that the dynamical systems theory approach, emphasizing the embodied embedded nature of cognition, might provide an important, non-representational alternative. We stress the importance of the challenge, raised by Andy Clark , to dynamical systems theory to deal with 'representation-hungry' cognitive tasks. We indicate a possible way to answer that challenge in a empirically applicable manner. We suggest that investigations of this kind strengthen a motto that can be used as an antidote to the traditional representational cravings of cognitive science: 'Don't use representations in explanation and modeling unless it is absolutely necessary.'
Discrete Thoughts: Why Cognition Must Use Discrete Representations
Mind & Language, 2003
Advocates of dynamic systems have suggested that higher mental processes are based on continuous representations. In order to evaluate this claim, we first define the concept of representation, and rigorously distinguish between discrete representations and continuous representations. We also explore two important bases of representational content. Then, we present seven arguments that discrete representations are necessary for any system that must discriminate between two or more states. It follows that higher mental processes require discrete representations. We also argue that discrete representations are more influenced by conceptual role than continuous representations. We end by arguing that the presence of discrete representations in cognitive systems entails that computationalism (i.e., the view that the mind is a computational device) is true, and that cognitive science should embrace representational pluralism.
1995
Recently, a new approach to modeling cognitive phenomena has been gaining recognition: the dynamical systems approach. Proponents of this theory claim to have identified a new paradigm for the study of cognition which is superior to both symbolicism and connectionism.
The presence of representations in a cognitive system has been a contentious issue since ages in the field of cognitive science. In this paper, I state in brief the various schools of thoughts that have come up with their own notion of representation. Further, we shall consider the Watt governor and its conception as a noncomputational, nonrepresentational model by Tim van Gelder. Such a view will be contrasted by William Bechtel’s argument with further reference to Hutto and Myin’s Radical Embodiment. Keywords: Representation, dynamical systems, classical computation
Cognitive Science, 2013
Mental representations have continuous as well as discrete, combinatorial properties. For example, while predominantly discrete, phonological representations also vary continuously; this is reflected by gradient effects in instrumental studies of speech production. Can an integrated theoretical framework address both aspects of structure? The framework we introduce here, Gradient Symbol Processing, characterizes the emergence of grammatical macrostructure from the Parallel Distributed Processing microstructure (McClelland & Rumelhart, 1986) of language processing. The mental representations that emerge, Distributed Symbol Systems, have both combinatorial and gradient structure. They are processed through Subsymbolic Optimization-Quantization, in which an optimization process favoring representations that satisfy well-formedness constraints operates in parallel with a distributed quantization process favoring discrete symbolic structures. We apply a particular instantiation of this framework, λ-Diffusion Theory, to phonological production. Simulations of the resulting model suggest that Gradient Symbol Processing offers a way to unify accounts of grammatical competence with both discrete and continuous patterns in language performance. The work discussed here was developed as one path for carrying out a research program that was already sketched by 1986 1 : (1) A PDP approach to cognitive macrostructure "The basic perspective of this book is that many of the constructs of macrolevel descriptions … can be viewed as emerging out of interactions of the microstructure of distributed models. … although we imagine that rule-based models of language acquisition … may all be more or less valid approximate macrostructural descriptions, 1 Important precedents include Hofstadter (1979, 1985). Other approaches to combining continuous activation spreading and symbolic structure, but without distributed representations (in the sense used here), include the ACT systems (Anderson & Lebiere, 1998), the LISA model (Hummel & Holyoak, 2003) and a range of hybrid architectures (Wermter & Sun, 2000). Gradient Symbol Systems There are also hybrid approaches which allow both discrete morphological and phonological symbolic representations to directly interact with continuous phonetic representations (e.g., phonetic exemplars; see Pierrehumbert, 2006, for a review). Our proposal maintains a clear separation between phonological and phonetic representations (allowing only the former to interact with morphological representations), and is unique in incorporating gradience within symbolic phonological representations, as well as within non-symbolic phonetic representations. We emphasize that we regard the GSPH truly as a working hypothesis. We would welcome future results showing how some of the structure that we must now simply assume to be in place can arise from learning, or results that show how to characterize, with some formal precision, a macrostructure that possesses the functional capabilities of our architecture but which deviates more significantly from symbolic theory. Work such as Plaut, McClelland, Seidenberg, & Patterson (1996) is quite promising on both accounts, and formal connections to the approach presented here would be extremely valuable. Our topic, the emergence of macrostructure, has been a main theme in the work of Jay McClelland. We view our approach as fundamentally consistent with his, but complementary. McClelland's approach assigns preeminent importance to gradience, with approximately discrete symbolic structure emerging in particular cognitive contexts. As discussed below, our view is that symbolic combinatorial structure provides a highly productive framework for developing theories of cognition. Our approach is therefore to start with systems utilizing discrete symbolic constituents and incorporate gradience as required by the data. We anticipate that the two approaches will eventually meet somewhere in a middle ground where the discrete and the continuous interact in a rich a constructive fashion.
Representation in dynamical and embodied cognition
Cognitive Systems Research, 2002
The move toward a dynamical and embodied understanding of cognitive processes initiated a debate about the usefulness of the notion of representation for cognitive science. The debate started when some proponents of a dynamical and embodied approach argued that the use of representations could be discarded in many circumstances. This remained a minority view, however, and there is now a tendency to shove this critique of the usefulness of representations aside as a non-issue for a dynamical and situated approach to cognition. In opposition, I will argue that the representation issue is far from settled, and instead forms the kernel of an important conceptual shift between traditional cognitive science and a dynamical and embodied approach. This will be done by making explicit the key features of representation in traditional cognitive science and by arguing that the representation-like entities that come to the fore in a dynamical and embodied approach are significantly different from the traditional notion of representation. This difference warrants a change of terminology to signal an important change in meaning.
Mental Representations and the Dynamic Theory of Mind
Logos & Episteme, 2012
In this paper I will investigate the possibility of defending the concept of 'mental representation' against certain contemporary critiques. Some authors, like Anthony Chemero, argue that it is possible to explain offline actions with dynamic concepts. Hence, the dynamic discourse preempts the representational one. I doubt that this is a recommendable strategy. A form of representation is necessary, though one which is different from the classical one. Instead of eliminating the concept of representation (as radical dynamicists do) or of splitting cognitive explanation in two separate discourses (as the adepts of the hybrid cognition version do), I consider that a dynamic concept of 'representation' is a better option. In my view, the higher level order resulted from the complex brain-body-environment coupling can be interpreted as being representational in nature. The dynamic paradigm involves a significant change concerning the intentional nature of representational states: the basic forms of representations are not maps of reality implemented as such in the brain, but limit conditions, attractors constraining the cognitive system's evolution in its space state to reach its goals. On a certain threshold of complexity, the system develops stable attractors and attractor landscapes which could be interpreted as standing for something outside the system. This conception offers the advantages of avoiding preemption argument, of unifying the cognitive explanation and, by its interscalar account, offers dynamic tools for building more complex artificial intelligent systems.