A set of hypotheses is formulated for a connectionist approach to cognitive modeling. These hypotheses are shown to be incompatible with the hypotheses underlying traditional cognitive models. The connectionist models considered are massively parallel numerical computational systems that are a kind of continuous dynamical system. The numerical variables in the system correspond semantically to fine-grained features below the level of the concepts consciously used to describe the task domain. The level of analysis is intermediate between those of symbolic cognitive models (...) and neural models. The explanations of behavior provided are like those traditional in the physical sciences, unlike the explanations provided by symbolic models. Higher-level analyses of these connectionist models reveal subtle relations to symbolic models. Parallel connectionist memory and linguistic processes are hypothesized to give rise to processes that are describable at a higher level as sequential rule application. At the lower level, computation has the character of massively parallel satisfaction of soft numerical constraints; at the higher level, this can lead to competence characterizable by hard rules. Performance will typically deviate from this competence since behavior is achieved not by interpreting hard rules but by satisfying soft constraints. The result is a picture in which traditional and connectionist theoretical constructs collaborate intimately to provide an understanding of cognition. (shrink)

In this brief space I want to describe how Chomsky's analysis of "psychological reality" departs from what I think is a fairly standard construal of the idea. This familiar formulation arises from distinguishing between someone's following a rule and someone's acting in conformity with a rule. The former idea, but not the latter, involves the idea that the person has some mental representation of the rule that plays a certain causal role in determining behavior. Although there may be many grammatical (...) rules to which a person's verbal behavior (ignoring slips of the tongue and other performance interferences) conforms, only some of those rules have psychological reality; these are the ones that are internalized in thought and play some appropriate causal role. (shrink)

A set of hypotheses is formulated for a connectionist approach to cognitive modeling. These hypotheses are shown to be incompatible with the hypotheses underlying traditional cognitive models. The connectionist models considered are massively parallel numerical computational systems that are a kind of continuous dynamical system. The numerical variables in the system correspond semantically to fine-grained features below the level of the concepts consciously used to describe the task domain. The level of analysis is intermediate between those of symbolic cognitive models (...) and neural models. The explanations of behavior provided are like those traditional in the physical sciences, unlike the explanations provided by symbolic models. (shrink)

Negative gegenseitige Befruchtung mag als seltsamer Titel erscheinen. Gegenseitige Befruchtung scheint per definitionem etwas Positives zu sein. Aber diese Untersuchung will die Aufmerksamkeit auf die Tatsache richten, daß in einer großen Anzahl von Fällen die gegenseitige Befruchtung einige negative Effekte auslöst; einige von ihnen mögen sogar katastrophal sein. Dieser Artikel kann die Aufmerksamkeit nur auf die negativen Auswirkungen und die Notwendigkeit lenken, die Bedingungen, welche den positiven oder negativen Ausgang oder die Unwirksamkeit einer bestimmten gegenseitigen Befruchtung determinieren, ausführlicher zu untersuchen; (...) um die Randbedingungen auf beiden Seiten zu formulieren: Das Befruchtende und das Befruchtete. (shrink)

This paper outlines some of the issues and basic philosophy that have guided my work and that of my students in the last ten years. It describes the progression of conceptual representational theories developed during that time, as well as some of the research models built to implement those theories. The paper concludes with a discussion of my most recent work in the area of modelling memory. It presents a theory of MOPs (Memory Organization Packets), which serve as both processors (...) and organizers of information in memory. This enables effective categorization of experiences in episodic memory, which in turn enables better predictive understanding of new experiences. (shrink)

This essay examines arguments for and against the proposition that Artificial Intelligence (AI) research makes an important contribution to the understanding of the human mind. A number of recent articles have seemed to question the value of Al ideas in specific domains (e.g., language. mental imagery, problem solving). In the present paper, it is argued that the real disagreement concerns the form of a scientific psychology. The critics of Artificial Intelligence believe that many acceptable psychological theories exist and the important (...) task of the scientist is to show that certain of these theories are true. They criticize AI research because it has not produced theories whose adequacy can be tested by empirical research. The supporters of AI research believe that no adequate psychological theories exist. They claim that the important task is to develop adequate theories and until such theories have been constructed, it is pointless to worry about empirical tests.A number of major arguments attacking and defending AI research are examined in order to illustrate that the real issue is methodological. It is concluded that the present debate is counterproductive since both sides assume that the methodological premise of the other side is wrong without attempting to refute it. A more constructive debate would address directly the more fundamental question of what methods offer the greatest promise of solving the major problems of psychology. (shrink)

Children learn their native language by exposure to their linguistic and communicative environment, but apparently without requiring that their mistakes be corrected. Such learning from “positive evidence” has been viewed as raising “logical” problems for language acquisition. In particular, without correction, how is the child to recover from conjecturing an over-general grammar, which will be consistent with any sentence that the child hears? There have been many proposals concerning how this “logical problem” can be dissolved. In this study, we review (...) recent formal results showing that the learner has sufficient data to learn successfully from positive evidence, if it favors the simplest encoding of the linguistic input. Results include the learnability of linguistic prediction, grammaticality judgments, language production, and form-meaning mappings. The simplicity approach can also be “scaled down” to analyze the learnability of specific linguistic constructions, and it is amenable to empirical testing as a framework for describing human language acquisition. (shrink)