 |
Britt Anderson
|
|
Research Interests I am interested in developing mechanistically specific models of cognitive phenomena and submitting those models to experimental examination. Currently, I am putting most of my energy on the topic of attention, broadly conceived. One interest is quasi-philosophical: what do we mean by the term ``attention'' and is our use consistent and cogent? In general, I feel attention is a "bad" word; attention is often reified. This reification leads to wasted experimental effort exploring nonsensical hypotheses and also obfuscates the presentation and interpretation of experimental data. What we need instead is an account of the experimental phenomena in terms of the experimental manipulations that lead to their being labeled as ``attentional.'' I think a good case can be made that attention is a Bayesian Decision Process (BDP). A BDP has a few core components: a prior probability distribution, a likelihood, and a cost function. From these, and an assumption of what is the measure of ``optimal,'' one can calculate the ideal solution. I conjecture that we will find that all our so-called attentional results can be explained in terms of these three factors and that we can do quite well without the term attention, except perhaps as a verbal shorthand for these sorts of effects. Therefore, I spend a fair amount of time working to test this idea psychophysically. I have been testing participants in an experiment where the prior probability of where targets can appear in a simple visual discrimination task is modulated. We have found a potent effect of spatial prior probability. Other follow up studies are on-going. A related notion of attention forks from the clinical syndrome of hemi-spatial neglect. This was an early research interest of mine during my years as an active practicing neurologist. From that experience, I came to appreciate that neglect is not homogeneous, strictly visual, nor simply spatial, and it certainly is not ``hemi;'' but still, in the colloquial sense, there does seem to be something common to many of these patients. My colleague, James Danckert, has been active in exploring the non-classical impairments of parietal lobe injury. Together, we are working on an account of parietal lobe injury that seeks to unify the mechanims under the concept of representational updating. Currently, we are working on a more theoretical paper that will make this notion of updating better defined and more concrete, at least from a functional standpoint. In addition, we are collaborating together and with our students on various psychophysical studies assessing how well participants with parietal lobe injury can learn to update an internal representation in response to environmental changes. One major research line has the participants playing Rock, Paper, Scissors. A second, looks at the ability to learn, covertly, a spatial probability distribution that influences target locations in a classification task. I am also cross appointed to the Centre for Theoretical Neuroscience (CTN) which brings together researchers from Biology, Mathematics, Philosophy and Psychology to develop biologically plausible computational models of neural systems. Students enrolled in a PhD in any of these departments can also take a Diploma in Theoretical Neuroscience through the CTN. If you are interested in graduate studies on these topics, drop me a line and we can see if we have common interests. Unlike many faculty in the department, I am happy to consider applications from individuals with non-psychology degrees, especially if they have computational training. Recent and Selected Publications
|
|
