Statistical properties of retrograde projections to macaque cortical areas V1 and V4

Abstract

Abstract: : Purpose: Injection of retrograde tracer in a cortical area reveals a unique constellation of areas, projecting to the given area. Do the numbers of neurons projecting from these areas to a target area display a unique connectivity profile? A previous study has claimed that small injections in the cortex generate patterns of connectivity that are over–dispersed and highly skewed, leading to low statistical power, and, thus, suggesting that the computation carried out by individual areas is either highly variable or does not require a constant input from each area [Scannell et al, Trans Roy Soc Lond (2000)]. This result could be the consequence of injection size. Methods: We analyzed the results generated by large injections of Fast blue and Diamidino Yellow to two visual areas (central V1 and V4) in 6 cynomologus monkeys. An exhaustive quantitative analysis was employed to estimate the number of neurons in each cortical area in order to explore the inter–individual pattern of connectivity. Results: Nineteen visual areas project to central V1 and 13 to V4. Analyses of these data reveal them to be over–dispersed but not significantly skewed. Power simulations for detecting differences between counts with these properties show that small differences can be detected with few samples given the numbers of counts observed per cortical area. The hypothesis that the profile of counts across cortical areas does not differ significantly across animals was evaluated using a Generalized Linear Model with either a Poisson or a Negative Binomial link. Only the latter properly accounted for the over–dispersion and did not reject the hypothesis. Slightly better results are obtained by treating animal as a random effect. Conclusions: These results show that neural count data reveal a signature profile of connectivity for a given target area when large individual injections are used. Failure to detect specific profiles with small injections could be due to the injection being restricted to a module within a cortical area.

Keywords: visual cortex • anatomy