Original project done for the course, MB208: Theoretical and Computational Neuroscience (Prof. Rishikesh Narayanan and Prof. SP Arun), at IISc, Bangalore.
It is the reductionist dream to explain the behavior of complex systems based on a set of well-defined simple principles. When it comes to neuroscience, one of the most prevalent reductionist approaches is to explain perceptual and behavioral observations based upon firing properties of single neurons. But for that to be made possible, it is essential to be able to quantify encoding properties at the level of single neurons. One of the common approaches, at least in vision science, is to display a set of carefully designed stimuli at the receptive field of a neuron of the visual areas, and then study the encoding properties of the neuron to those stimuli. Though this is a reasonable approach and have revealed crucial properties of visual processing in the brain, there might be certain problems with what is defined to be the receptive field of a neuron, and the method used to localize it spatially. There has been ample work that has shown that the firing of visual neurons gets affected by stimulus presented outside their classical receptive fields (Zhou et. al, 2000). In this study, we tried to understand computationally, how the Bayesian decoding of stimulus presented to a neuron's receptive field can be affected by the influence of the area surrounding the receptive field. We took the case of orientation-selective neurons of V1 for simplicity and inspected how accurately the orientation of the bar presented to its receptive field can be decoded from its response when other oriented bars were also presented close to the receptive field of the neuron. We also tried to estimate if there is an optimal amount of 'influence' from the area surrounding the receptive field for better Bayesian decoding of the presented stimulus.
Original project done for the course, CH242: Special Topics in theoretical Biology (Prof. Narendra Dixit), at IISc, Bangalore.
The survival and success of an individual depends upon its efficiency in foraging and finding mates. Many different strategies have evolved to do the latter, varying from one copulation followed by death, to one male copulating with ≈ 200 females every breeding season. In species that practice polygamy, sexual selection is very strong and hence it is the strong, large and aggressive males that get to mate with the most females. Dominance of strong males in animal societies have given rise to a social and reproductive structure called a ‘Harem’, which basically is a collection of multiple females with a strong (central) male that mates with them and protects them. The central male maintains small harems alone, as harem size increases, he is unable to fend off outside males (who do not have females of their own) from sneaking up on his harem and mating with the females. The central male thus recruits an outsider to share the sentry duty and as a reward, he allows him (now called a peripheral male) partial mating rights. The success of various mating strategies (including harems) as a function of certain general parameters is a problem that has been theoretically addressed in the literature. However, the specific biological details of a particular mating system (like the central, peripheral, outsider hierarchy in the case of a harem) has never been used in constructing a model describing it. Of the different mating systems, harems proved the most interesting to us as the sex ratio within a harem seemed close to that predicted by the ‘Dairy Farmer Optimum’. Additionally, the presence of a large proportion of reproductively capable males without mating rights was another reason why harems seemed an interesting system to study and we atried to address the following questions in this study: