Online research in philosophy

Computers today are not only the calculation tools - they are directly (inter)acting in the physical world which itself may be conceived of as the universal computer (Zuse, Fredkin, Wolfram, Chaitin, Lloyd). In expanding its domains from abstract logical symbol manipulation to physical embedded and networked devices, computing goes beyond Church-Turing limit (Copeland, Siegelman, Burgin, Schachter). Computational processes are distributed, reactive, interactive, agent-based and concurrent. The main criterion of success of computation is not its termination, but the adequacy of its response, its speed, generality and flexibility; adaptability, and tolerance to noise, error,faults, and damage. Interactive computing is a generalization of Turing computing, and it calls for new conceptualizations (Goldin, Wegner). In the info-computationalist framework, with computation seen as information processing, natural computation appears as the most suitable paradigm of computation and information semantics requires logical pluralism.

The main claim of this paper is that notions of implementation based on an isomorphic correspondence between physical and computational states are not tenable. Rather, ``implementation'' has to be based on the notion of ``bisimulation'' in order to be able to block unwanted implementation results and incorporate intuitions from computational practice. A formal definition of implementation is suggested, which satisfies theoretical and practical requirements and may also be used to make the functionalist notion of ``physical realization'' precise. The upshot of this new definition of implementation is that implementation cannot distinguish isomorphic bisimilar from non-isomporphic bisimilar systems anymore, thus driving a wedge between the notions of causal and computational complexity. While computationalism does not seem to be affected by this result, the consequences for functionalism are not clear and need further investigations.

The central paradigm of arti?cial intelligence is rapidly shifting toward biological models for both robotic devices and systems performing such critical tasks as network management, vehicle navigation, and process control. Here we use a recent mathematical analysis of the necessary conditions for consciousness in humans to explore likely failure modes inherent to a broad class of biologically inspired computing machines. Analogs to developmental psychopathology, in which regulatory mechanisms for consciousness fail progressively and subtly understress, and toinattentional blindness, where a narrow 'syntactic band pass' de?ned by the rate distortion manifold of conscious attention results in pathological ?xation, seem inevitable. Similar problems are likely to confront other possible architectures, although their mathematical description may be far less straightforward. Computing devices constructed on biological paradigms will inevitably lack the elaborate, but poorly understood, system of control mechanisms which has evolved over the last few hundred million years to stabilize consciousness in higher animals. This will make such machines prone to insidious degradation, and, ultimately, catastrophic failure.

After briefly discussing the relevance of the notions computation and implementation for cognitive science, I summarize some of the problems that have been found in their most common interpretations. In particular, I argue that standard notions of computation together with a state-to-state correspondence view of implementation cannot overcome difficulties posed by Putnam's Realization Theorem and that, therefore, a different approach to implementation is required. The notion realization of a function, developed out of physical theories, is then introduced as a replacement for the notional pair computation-implementation. After gradual refinement, taking practical constraints into account, this notion gives rise to the notion digital system which singles out physical systems that could be actually used, and possibly even built.

Cognitive science has been dominated by the computational conception that cognition is computation across representations. To the extent to which cognition as computation across representations is supposed to be a purposive, meaningful, algorithmic, problem-solving activity, however, computers appear to be incapable of cognition. They are devices that can facilitate computations on the basis of semantic grounding relations as special kinds of signs. Even their algorithmic, problem-solving character arises from their interpretation by human users. Strictly speaking, computers as such — apart from human users — are not only incapable of cognition, but even incapable of computation, properly construed. If we want to understand the nature of thought, then we have to study thinking, not computing, because they are not the same thing.

Computational models are abstract entities that physical systems can implement or realize. Structuralism about computational implementation, espoused by Chalmers and others, holds that a physical system realizes a computational model just in the case the system exhibits a pattern of causal organization isomorphic to the model’s formal structure. I argue against structuralism through counter-examples drawn from computer science. On my opposing view, computational implementation sometimes requires instantiating semantic properties that outstrip any pattern of causal organization specified by the formal model. In developing my argument, I defend anti-individualism about computational implementation. More specifically, I argue that relations to the social environment sometimes help determine whether a physical system realizes a computational model.

Currently, there is widespread skepticism that higher cognitive processes, given their apparent flexibility and globality, could be carried out by specialized computational devices, or modules. This skepticism is largely due to Fodor’s influential definition of modularity. From the rather flexible catalogue of possible modular features that Fodor originally proposed has emerged a widely held notion of modules as rigid, informationally encapsulated devices that accept highly local inputs and whose opera- tions are insensitive to context. It is a mistake, however, to equate such features with computational devices in general and therefore to assume, as Fodor does, that higher cognitive processes must be non-computational. Of the many possible non-Fodorean architectures, one is explored here that offers possible solutions to computational problems faced by conventional modular systems: an ‘enzymatic’ architecture. Enzymes are computational devices that use lock-and-key template matching to iden- tify relevant information (substrates), which is then operated upon and returned to a common pool for possible processing by other devices. Highly specialized enzymes can operate together in a common pool of information that is not pre-sorted by information type. Moreover, enzymes can use molecular ‘tags’ to regulate the operations of other devices and to change how particular substrates are construed and operated upon, allowing for highly interactive, context-specific processing. This model shows how specialized, modular processing can occur in an open system, and suggests that skepti- cism about modularity may largely be due to failure to consider alternatives to the standard model.

Currently, there is widespread skepticism that higher cognitive processes, given their apparent flexibility and globality, could be carried out by specialized computational devices, or modules. This skepticism is largely due to Fodor’s influential definition of modularity. From the rather flexible catalogue of possible modular features that Fodor originally proposed has emerged a widely held notion of modules as rigid, informationally encapsulated devices that accept highly local inputs and whose opera- tions are insensitive to context. It is a mistake, however, to equate such features with computational devices in general and therefore to assume, as Fodor does, that higher cognitive processes must be non-computational. Of the many possible non-Fodorean architectures, one is explored here that offers possible solutions to computational problems faced by conventional modular systems: an ‘enzymatic’ architecture. Enzymes are computational devices that use lock-and-key template matching to iden- tify relevant information (substrates), which is then operated upon and returned to a common pool for possible processing by other devices. Highly specialized enzymes can operate together in a common pool of information that is not pre-sorted by information type. Moreover, enzymes can use molecular ‘tags’ to regulate the operations of other devices and to change how particular substrates are construed and operated upon, allowing for highly interactive, context-specific processing. This model shows how specialized, modular processing can occur in an open system, and suggests that skepti- cism about modularity may largely be due to failure to consider alternatives to the standard model.

In both the search for ever smaller and faster computational devices, and the search for a computational understanding of biological systems such as the brain, one is naturally led to consider the possibility of computational devices the size of cells, of molecules, of atoms, or even the size of sub-atomicquanta. Thus, it should not be surprising to nd that the idea ofquantum computation is not new; in particular, Deutsch has published striking papers on the notion as far back as 1985; and there were speculations on the issue even before that.

The increased interactivity and connectivity of computational devices along with the spreading of computational tools and computational thinking across the fields, has changed our understanding of the nature of computing. In the course of this development computing models have been extended from the initial abstract symbol manipulating mechanisms of stand-alone, discrete sequential machines, to the models of natural computing in the physical world, generally concurrent asynchronous processes capable of modelling living systems, their informational structures and dynamics on both symbolic and sub-symbolic information processing levels. Present account of models of computation highlights several topics of importance for the development of new understanding of computing and its role: natural computation and the relationship between the model and physical implementation, interactivity as fundamental for computational modelling of concurrent information processing systems such as living organisms and their networks, and the new developments in logic needed to support this generalized framework. Computing understood as information processing is closely related to natural sciences; it helps us recognize connections between sciences, and provides a unified approach for modeling and simulating of both living and non-living systems.