“School of Cognitive Sciences”

Back to Papers Home
Back to Papers of School of Cognitive Sciences

Paper   IPM / Cognitive Sciences / 12255
School of Cognitive Sciences
  Title:   Decoding the spatial attention state from local field potentials in area MT of macaque monkeys
  Author(s): 
1.  Mohammad Reza Daliri
2.  M. Sarraf
3.  S. TREUE
  Status:   In Proceedings
  Proceeding: Neuroscience, 2011
  Year:  2011
  Supported by:  IPM
  Abstract:
Attention has been shown to modulate neuronal activity in many areas of cortex in man and other primates. This modulation includes changes in gamma band activity of local field potentials (LFP). In the current study we investigated whether different attentional states can be discerned based on LFP activity. We recorded action potentials of individual neurons and LFPs from area MT in the visual cortex of a male macaque monkey while the animal was performing an attentional task. The monkey had to fixate on a central fixation point and was instructed to direct his attention onto one of two random dot patterns (RDP) presented simultaneously on a computer screen. The behaviorally relevant stimulus, the target, was specified using a cue at the beginning of each trial. The two RDPs were moved at the same direction (eight possible directions) at the preferred speed of the neuron specified using the spike tuning functions. One of them was placed inside the receptive field and the other outside. We analyzed the LFPs starting from the stimulus presentation till 680 ms after stimulus presentation. The signals were band-pass filtered with Butterworth IIR filter with cut-off frequencies of 1Hz and 100 Hz for eliminating the DC offset and higher frequencies, and it was stop band-filtered with an IIR filter of the range 49-51 Hz at 3dB in order to suppress the energy of 50 Hz power line noise. For all filtered LFPs, continuous wavelet analysis is performed to compute the scalogram of wavelet coefficients using Morlet wavelet. In order to find the decoding mechanism of attentional states, the time-scale power features is segmented with the windows of 1 scale and the time interval of 100ms with 10ms overlapping. Then SVM classifier was applied on 50The results demonstrate that the best decoding performances were achieved in frequencies above 30 Hz bands over the time period of 220-525ms after stimulus onset (p<0.001). We also found that in lower frequency bands (<20Hz), the decoding of attentional state can be done approximately after 440ms after stimulus onset (p<0.001). The results indicate that it is possible to decode the attentional state of the animal using the gamma band of the local field potentials. This is compatible with the previous findings that show attention enhances the gamma band activity of the LFPs in different brain areas.

Download TeX format
back to top
scroll left or right