Abstract
To examine the effect of attention on the synchronization of the LFP (local field potential) of V4 neurons, Fries et al. (2001) calculated STAs (spike trigerred averages). The power spectra of the STAs show which frequency is dominant in the oscillation of the LFP. Attention increased and decreased STA power in the gamma-band (35–90Hz) and low-band (<10Hz), respectively. Based on the evidence they concluded that attention modulates synchronization between V4 neurons. This study shows that attentional modulation on STAs can be explained by a rate-based attentional modulation without directly controlling synchronization. Our model has a hierarchical structure corresponding to V1, V2 and V4. Each neuron has broadly tuned feature selectivity, and there is competition among neurons with different feature preference. The LFP is computed by summing the activity of neighboring V4 neurons, and the LFP shows oscillation because of overlapping receptive fields. The dominant frequency in the power spectrum of the STAs depends on spikes from V2 neurons. The components of the STAs in gamma- and low-band originate from spiking V2 neurons with identical and different feature preference, respectively. In our model, attention multiplicatively increases the feature selectivity of V2 neurons. When a high-contrast stimulus in the receptive field is attended, the rate of V2 neurons preferring the presented stimulus does not change compared to that without attention because of upper firing rate limit. On the contrary, attention increases the rate of V2 neurons with slightly different preferences than the presented stimulus up to gamma-frequency. As a result, V4 neurons receive more spikes in the gamma-band, and power spectra of STAs in the gamma-band are increased by attention. Our rate-based model simulated the attentional modulation on STAs power spectra. The result suggests that attentional modulation on STA power may not indicate the existence of synchronization-based attention.
Support from 21st Century COE (D-2 to Kyoto Univ.), Advanced and Innovational Research program in Life Sciences from JMEXT, and PRESTO from JST.