Reader Comments

Post a new comment on this article

The dependence of directional selectivity on input frequency

Posted by Justin_Harrris on 19 Jan 2007 at 06:17 GMT

In this paper, Puccini, Compte and Maravall show that, using a simple integrate-and-fire model incorporating directional tuning of synaptic inputs, directional selectivity in barrel cortex neurons is present for low frequency inputs but is lost at high frequency. This nicely confirms electrophysiological evidence for a similar relationship between frequency and direction selectivity in barrel cortex (1). To locate this finding in a behavioural context, Puccini et al suggest that low frequency whisker deflections are likely to carry information about the position of an object contacted by the whisker, while high frequency changes in whisker movement would be generated by surface texture.

Carrying this idea further, we speculate that there may be a meaningful advantage to the putative dissociation between direction-selectivity in cortical neurons and the coding of macrostructural vs microstructural features of an object. Specifically, the loss of directional tuning with high frequency input may preserve the capacity of those same neurons to code directional information about macrostructural features such as edge orientation or surface contour. Let us consider the alternative – that directional tuning is maintained at high frequency input – in a situation where the rat’s whisker hits and sweeps across the surface of an object. The initial coding of relevant macrostructural information would become integrated with high-frequency input generated by the movement of the whisker over the surface of the object. If one assumes that, at least for natural objects, surface textures have spatially random structure, then coding of this directional information would have little value and may only serve to add noise to the more meaningful directional coding of macrostructural information. As such, the loss of directional tuning with high frequency input may facilitate the extraction and retention of relevant information about an object or surface.

Justin Harris & Ehsan Arabzadeh
School of Psychology
The University of Sydney
Australia

(1) Arabzadeh E, Zorzin E, Diamond ME. (2005) Neuronal encoding of texture in the whisker sensory pathway. PLoS Biol 3: e17. doi:10.1371/journal.pbio.0030017

RE: The dependence of directional selectivity on input frequency

mmaravall replied to Justin_Harrris on 29 Jan 2007 at 14:48 GMT

I agree that this is a very interesting possibility. Harris and Arabzadeh point out something that could be an interesting consequence of the results in the paper but which we could only touch on in our Discussion, namely that the specific information that responses encode about stimulus features may change in time, so that the code underlying object representations may in fact be quite “fluid” and modulated within a short timescale, representing different properties of the stimulus.