Précis of “ Modeling visual lateralization and interhemispheric communication ”

Lateralization is intertwined with virtually every function that we think makes us human, including language, fine motor skills, and visuo-spatial processing. In my PhD thesis, I attempted to give a computational account of lateralization in visual processing, offering hypotheses about its neural underpinnings, neurodevelopmental origins, and relationship to interhemispheric transfer. To do so, I performed new analyses on existing models, used them to account for new data, and extended them to model developmental processes. The result is a computational model of lateralization in visual processing, new models and ideas about interhemispheric transfer across species (including humans), and a number of quantitative and qualitative neuroanatomical predictions. The thesis is separated into three main projects, each focused on data from a different discipline. In the first project (Chapters 2-4), I used neural networks to model a key type of connection in visual processing, long-range lateral connections in cortex. In Chapters 2 and 3, I examined how variations in the anatomy of these connections leads to spatial frequency biases in the model, which lead to processing differences that mirror those found in behavioral experiments using visual hemifield presentation of hierarchical letters, faces, spatial frequency gratings. Finally, in Chapter 4, I showed that well-known developmental constraints can cause this connection asymmetry in the model. These hemispheric models were not connected. In preparation for modeling hemispheric interactions between the hemispheres, I followed up with two projects, each examining influential papers about how anatomical and physiological properties of the corpus callosum affect interhemispheric interactions. The first of these projects (Chapter 5) focused on modeling results that suggest that larger communication delays across callosal connections weaken interhemispheric communication (Ringo, Doty, Demeter, & Simard, 1994). I reanalyzed the neural network results, and found that delay magnitude does not change interhemispheric communication it only delays it! I then used a similar neural network to show that conduction delay variability can decrease interhemispheric communication, and argued that this variability is present early in development due to biophysical properties of immature connections. The final project (Chapter 6) focused on comparing intrahemispheric and interhemispheric connection fiber counts across largeand small-brained species, using allometric regression (Rilling & Insel, 1999), a technique common to biological anthropology. The original paper concluded that the size of the corpus callosum is decreasing on an evolutionary time scale. I expected that their conclusion was wrong, and that there would be no decrease with a more accurate analysis. Through a careful analysis of the data in the literature, introducing several novel corrections to the data, I found to my surprise that they underestimated the reduction. In understanding the significance of this large reduction, I discovered a new way to think about the data in terms of functional connections, which I think is more relevant to the brain’s processing. I showed that this new measure explains Rilling & Insel’s results, and scales perfectly (with a slope of 1) across mammals. I concluded that the relative size of the functional connectivity of the corpus callosum is not changing with brain size.