New insights into sensory representations-Lessons from the auditory cortex

In the brain, there are highly ordered representations of sensory input. The existence of orientation columns in the visual cortex where columns of neurons situated next to each other respond to slightly different stimulus orientations and the barrel cortex in S1 where each barrel faithfully receives inputs from one whisker are testimony to this. Recently two papers in the same issue of Nature Neuroscience dealt with the fidelity of sensory representations in the auditory cortex. Rothschild et al. and Bandyopadhyay et al. used in-vivo two photon microscopy to map tone evoked activity in the primary auditory cortex (A1). This is done by bulk loading a large area of the cortex with a membrane permeable calcium dye. When a neuron that has taken up the dye fires an action potential, there is also a transient influx of calcium (related to synaptic transmission). The interaction of the calcium ions with the calcium indicator can be visualized under a two-photon microscope. 

Both studies showed, that unlike the visual and barrel cortices, the auditory cortex appears to have a tonotopic map that is fractured. Firstly, less than half the neurons were responsive, and even if they did respond, neurons with similar tuning curves were as likely to be located next to each other as neurons with very different tuning curves. Fig 1 (taken from a review by Castro and Kandler, Nature Neuroscience 2010) elaborates on this.

Fig1a would be the classic tonotopic map where there is a smooth change in frequencies along the rostrocaudal axis with more rostral neurons coding for higher frequencies while more caudal neurons code for lower ones. Here, tonotopy is maintained on the local as well as the global scale. Fig 1b summarizes the results of Rothschild et al. and Bandyopadhyay et al. Although the tonotopic organisation is maintained on a more global scale, locally the map appears to be fractured (From Castro and Kandler, 2010)

So why is the tonotopic map fractured? One possibility is that the thalamocortical projections from the auditory thalamus to A1 become scattered en route. Alternately, thalamocortical axons may be arranged tonotopically but resulting intra-cortical processing may result in the fractured nature of A1. To distinguish between these two possibilities, Bandyopadhyay et al labeled cells with two different calcium indicators, Fluo 4, a low affinity indicator that responds only to spikes in the cells, and OGB-1,  a high affinity indicator that responds to subthreshold synaptic inputs into the cells. They found that subthreshold maps were more ordered in comparison to suprathreshold maps based on spiking (Fig 2).

Fig2; Subthreshold and Suprathreshold maps (From Castro and Kandler, 2010)

So what does all of this mean? Bandyopadhyay et al suggest that although two neighboring neurons may receive similar , correlated inputs, they may be part of different fine-scaled assemblies that could process inputs differently. Two adjacent cells may be selective for different input features or different stimulus attributes. Taken together, both studies indicate that frequency is perhaps not the most important feature coded by A1 neurons. Neither is intensity tuning or bandwidth. A1 neurons probably respond to meaningful stimuli and not just to simple sound parameters. This is supported by studies that have shown that A1 neurons are best driven by spectrally and temporally rich stimuli.

References:

Castro JB, & Kandler K (2010). Changing tune in auditory cortex. Nature neuroscience, 13 (3), 271-3 PMID: 20177415

Rothschild, G., Nelken, I., & Mizrahi, A. (2010). Functional organization and population dynamics in the mouse primary auditory cortex Nature Neuroscience, 13 (3), 353-360 DOI: 10.1038/nn.2484

Bandyopadhyay S, Shamma SA, & Kanold PO (2010). Dichotomy of functional organization in the mouse auditory cortex. Nature neuroscience, 13 (3), 361-8 PMID: 20118924