Although much has been written on this subject, there still seems to be considerable confusion in the literature concerning dissociations, double dissociations and what they really mean, especially when connectionist or neural network models are involved. In this paper I attempt to clarify matters by looking at the subject from the point of view of patterns of learning rates in neural network models.

A considerable body of evidence from prosopagnosia, a deficit in face recognition dissociable from nonface object recognition, indicates that the visual system devotes a specialized functional area to mechanisms appropriate for face processing. We present a modular neural network composed of two “expert ” networks and one mediating “gate ” network with the task of learning to recognize the faces of 12 individuals and classifying 36 nonface objects as members of one of three classes. While learning the task, the network tends to divide labor between the two expert modules, with one expert specializing in face processing and the other specializing in nonface object processing. After training, we observe the network's performance on a test set as one of the experts is progressively damaged. The results roughly agree with data reported for prosopagnosic patients: as damage to the “face ” expert increases, the network's face recognition performance decreases dramatically while its object classification performance drops slowly. We conclude that data-driven competitive learning between two unbiased functional units can give rise to localized face processing, and that selective damage in such a system could underlie prosopagnosia.

A key claim of current theoretical analyses of the memory impairments associated with amnesia is that certain distinct forms of learning and memory are spared. A compelling example is that amnesic patients and controls are indistinguishable in repetition priming but amnesic patients are impaired at recognizing the study items. The authors show that this pattern of results is predicted by a single-system connectionist model of learning in which amnesia is simulated by a reduced learning rate. They also demonstrate that the model can reproduce the converse pattern in which priming but not recognition is impaired if the input is assumed to be additionally degraded in a priming test. The authors conclude that dissociations between priming and recognition do not require functionally or neurally distinct memory systems. According to an influential view, memory is not a unitary faculty but is composed of multiple systems that work independently of each other (Gabrieli, 1998; Squire, 1994). The most prominent distinction that has been proposed is between declarative and nondeclarative memory. Declarative (or explicit) memory is usually characterized by the conscious and intentional recollection of knowledge. Typical tests of declarative memory involve

A long-standing debate regarding the representation of semantic knowledge is whether such knowledge is represented in a single, amodal system or whether it is organised into multiple subsystems based on modality of input or type of information. The current paper presents a distributed connectionist model of semantics that constitutes a middle ground between these unitary- versus multiple-semantics accounts. In the model, semantic representations develop under the pressure of learning to mediate between multiple input and output modalities in performing various tasks. The system has a topographic bias on learning that favours short connections, leading to a graded degree of modality-specific functional specialisation within semantics. The model is applied to the specific empirical phenomena of optic aphasia—a neuropsychological disorder in which patients exhibit a selective deficit in naming visually presented objects that is not attributable to more generalised impairments in object recognition (visual agnosia) or naming (anomia). As a result of the topographic bias in the model, as well as the relative degrees of systematicity among tasks, damage to connections from vision to regions of semantics near phonology impairs visual object naming far more than visual gesturing or tactile naming, as observed in optic aphasia. Moreover, as in optic aphasia, the system is better at generating the name of an action associated with an object than at generating the name of the object itself, because action naming receives interactive support from the activation of action representations. The ability of the model to account for the pattern of performance observed in optic aphasia across the full range of severity of impairment provides support for the claim that semantic representations exhibit graded functional specialisation rather than being entirely amodal or modality-specific.

Abstract: Modularity in the human brain remains a controversial issue, with disagreement over the nature of the modules that exist, and why, when and how they emerge. It is a natural assumption that modularity offers some form of computational advantage, and hence evolution by natural selection has translated those advantages into the kind of modular neural structures familiar to cognitive scientists. However, simulations of the evolution of simplified neural systems have shown that, in many cases, it is actually non-modular architectures that are most efficient. In this paper, the relevant issues are discussed and a series of simulations are presented that reveal crucial dependencies on the details of the learning algorithms and tasks that are being modelled, and the importance of taking into account known physical brain constraints, such as the degree of neural connectivity. A pattern is established which provides one explanation of why modularity should emerge reliably across a range of neural processing tasks.

There is a considerable degree of modularity in the human brain and numerous reasons why it might have evolved to be that way. One may be tempted to build such modularity into our artificial systems. Indeed, Rueckl, Cave & Kosslyn (1989) demonstrated how a clear advantage in having a modular architecture can exist in neural network models. Here I present a series of simulations of the evolution of such neural systems that show how the advantage can cause modularity to evolve. However, I shall also show that it is possible to evolve even more efficient systems that are not modular. It seems that making the decision whether to build modularity into our systems is not as easy as is often thought.

Abstract not found

Researchers interested in human cognitive processes have long used computer simulations to try to identify the principles of cognition. The strategy has been to build computational models that embody putative principles and then to examine how well such models capture human performance in cognitive tasks. Until the 1980’s, this effort was undertaken within the context of the “computer metaphor” of mind. Researchers built computational models based on the conceptualization that the human mind operated as though it were a conventional digital computer. However, with the advent of so-called connectionist, neural network, or parallel distributed processing models (Anderson,

this paper, our goals are to introduce and to discuss these issues. We argue for an essentially utilitarian view of computational modeling. We suggest that the main function of computational modeling is to support an interactive process of "probing and prediction" through which models can be interacted with in a way that provides both guidance for empirical research and also sufficient depth to support interactive modification of the underlying theory. We propose that models, just as the systems they are models of, can only be understood (and evaluated) with respect to a given level of description and a specific set of criteria associated with that level. We also claim that models gain explanatory power as well as practical usefulness when they are emergent, that is, when they provide an account of how the principles of organization at a given level of description constrain and define structure at a higher level of description. For this reason, connectionist models appear to provide the most fruitful modeling framework today.

Abstract not found