We consider the problem of zero-shot learning, where the goal is to learn a classifier f: X → Y that must predict novel values of Y that were omitted from the training set. To achieve this, we define the notion of a semantic output code classifier (SOC) which utilizes a knowledge base of semantic properties of Y to extrapolate to novel classes. We provide a formalism for this type of classifier and study its theoretical properties in a PAC framework, showing conditions under which the classifier can accurately predict novel classes. As a case study, we build a SOC classifier for a neural decoding task and show that it can often predict words that people are thinking about from functional magnetic resonance images (fMRI) of their neural activity, even without training examples for those words. 1

This article presents a sobering view of the discipline of cognitive neuropsychology as practised over the last three or four decades. Our judgement is that, although the study of abnormal cognition resulting from brain injury or disease in previously normal adults has produced a catalogue of fascinating and highly selective deficits, it has yielded relatively little advance in understanding how the brain accomplishes its cognitive business. We question the wisdom of the following three ‘choices ’ in mainstream cognitive neuropsychology: (a) single-case methodology, (b) dissociation between functions as the most important source of evidence, and (c) a central goal of diagramming the functional architecture of cognition rather than specifying how its components work. These choices may all stem from an excessive commitment to strict and fine-grained modularity. Although different brain regions are undoubtedly specialised for different functions, we argue that parallel and interactive processing is a better assumption about cognitive processing. The essential goal of specifying representations and processes can, we claim, be significantly assisted by

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.

Representations of linguistic information and the neural substrates that underlie them are incredibly complex. This chapter illustrates how connectionist modeling has furthered our understanding of normal and impaired processing in three related domains – semantic memory, knowledge of grammatical class, and word reading – and how the

Thirty years ago, grounded cognition had roots in philosophy, perception, cognitive linguistics, psycholinguistics, cognitive psychology, and cognitive neuropsychology. During the next 20 years, grounded cognition continued developing in these areas, and it also took new forms in robotics, cognitive ecology, cognitive neuroscience, and developmental psychology. In the past 10 years, research on grounded cognition has grown rapidly, especially in cognitive neuroscience, social neuroscience, cognitive psychology, social psychology, and developmental psychology. Currently, grounded cognition appears to be achieving increased acceptance throughout cognitive science, shifting from relatively minor status to increasing importance. Nevertheless, researchers wonder whether grounded mechanisms lie at the heart of the cognitive system or are peripheral to classic symbolic mechanisms. Although grounded cognition is currently dominated by demonstration experiments in the absence of well-developed theories, the area is likely to become increasingly theory driven over the next 30 years. Another likely development is the increased incorporation of grounding mechanisms into cognitive architectures and into accounts of classic cognitive phenomena. As this incorporation occurs, much functionality of these architectures and

Forthcoming in The Philosophical Quarterly. To be reprinted in Macpherson, F. and Hawley, K.

Abstract not found

A key issue that continues to generate controversy concerns the nature of the psychological, computational, and neural mechanisms that support the visual recognition of objects such as faces and words. While some researchers claim that visual recognition is accomplished by category-specific modules dedicated to processing distinct object classes, other researchers have argued for a more distributed system with only partially specialized cortical regions. Considerable evidence from both functional neuroimaging and neuropsychology would seem to favour the modular view, and yet close examination of those data reveals rather graded patterns of specialization that support a more distributed account. This paper explores a theoretical middle ground in which the functional specialization of brain regions arises from general principles and constraints on neural representation and learning that operate throughout cortex but that nonetheless have distinct implications for different classes of stimuli. The account is supported by a computational simulation, in the form of an artificial neural network, that illustrates how cooperative and competitive interactions in the formation of neural representations for faces and words account for both their shared and distinctive properties. We set out a series of empirical predictions, which are also examined, and consider the further implications of this account.

As will be obvious from the responses to it in this issue of Cognitive Neuropsychology, the paper by Myrna Schwartz and Gary Dell ― on the uses and abuses of case-series methodology in cognitive neuropsychology ― has appropriately attracted a lot of interest. We endorse a great deal of the content of the Schwartz and Dell article (hereinafter, S&D), and will describe these sources of agreement below, although many have been noted in previous discussions of this topic (e.g., Lambon Ralph, Moriarty, & Sage, 2002, Patterson & Plaut, 2009). We also have a few, mainly minor, questions or quibbles in connection with some of these nods of our heads; these will also be addressed below. It seems appropriate, however, to start by nailing our colours to the mast. All researchers presumably realise that there is no one correct or best methodology, and further that ― even if there were ― it would probably not be feasible to apply it to every research question. Thus we know that a case-series approach in neuropsychology is not always possible, and indeed all three of us have performed single-case studies, either by choice or by necessity. If we have

Finite case series or infinite single-case studies?