My main research interests are semantic / conceptual knowledge and language. I am interested in how semantic and lexical representations are acquired with development as well as how they deteriorate upon brain damage or degeneration. The focus of my research is - through neural network modeling - to show how input of different types (perceptual / linguistic / motor) and from different modalities (visual / auditory / tactile) is integrated into cross-modal "semantic" representations. An important aspect of my work is the notion that lexical information is not represented separately; it is integrated with other kinds of information and becomes a part of conceptual knowledge. One line of research explores the classic Whorfian hypothesis that the language one speaks affects the way they perceive, reason about, and act in the world. Using a neural network, we sought to provide a mechanistic account of how that may happen in the brain: paper abstract: How do the languages we speak shape the way we think? In a series of studies, Boroditsky, Schmidt, and Phillips (2003) investigated the effect of grammatical gender on people’s responses to questions about the properties and similarity relations among objects. Here, we use a connectionist network to simulate these findings and find a possible mechanism for linguistic relativity effects (such as effects of grammatical categorization). The model’s behavior paralleled the effects seen in the human data. The results also suggest that the within- and between- category similarity relations among objects may play a role in generating these effects. Another issue that I find close to heart has to do with individual differences and their importance in interpreting and accounting for behavioural data, especially in neuropsychology. The notion of individual differences is largely ignored in cognitive neuropsychology. In my view, many paradoxical findings from individual patients or very small groups of patients that give rise to theoretical debates about the organizational and/or representational structure of the brain can be explained if viewed within a larger sample of data. In a recent paper, we presented a neural network model of semantic and lexical deficits in semantic dementia, where we explored the notion of individual differences: paper abstract: In semantic dementia (SD), there is a correlation between performance on semantic tasks such as picture naming and lexical tasks such as reading aloud. However, there have been a few case reports of patients with spared reading despite profound semantic impairment. These reports have sparked an ongoing debate about how the brain processes conceptual versus lexical knowledge. One possibility is that there are two functionally distinct systems in the brain - one for semantic and one for lexical processing. Alternatively, there may be a single system involved in both. We present a computational investigation of the role of individual differences in explaining the relationship between naming and reading performance in five SD patients, among whom there are cases of both association and dissociation of deficits. We used a connectionist model where information from different modalities feeds into a single integrative layer. Our simulations successfully produced the overall relationship between reading and naming seen in SD and provided multiple fits for both association and dissociation data, suggesting that a single, cross-modal, integrative system is sufficient for both semantic and lexical tasks and that individual differences among patients are essential in accounting for variability in performance. An extensive review of theoretical views of the structure and organization of conceptual knowledge, and relevant findings from semantic dementia can be found here. My dissertation project builds on my previous modeling work of semantic dementia, and will further advance the notion that conceptual knowledge and lexical knowledge are subserved by a common underlying cognitive system.