Biomedical Engineering Reference
In-Depth Information
Another pair of neurons (sig040a dmPFC and sig030a MC)
in different cortical areas has an opposite pattern. The neu-
rons do not share components, and the cross-correlation, rate-
correlation, and trial-by-trial correlation of these neurons is small;
however, we find strong JPSTH correlations, predictive inter-
actions, and redundant network interactions. This pair is easily
ignored because of a lack of clear cross-correlations or trial-by-
trial correlations; nonetheless, this pair has strong JPSTH and
predictive relationships. These neurons likely have a strong
signal
(where signal is controlling response to the stimulus) correlation
(38)
, in that they seem to be working together to predict behav-
ior. Their functional interactions may allude to top-down control
of dmPFC on motor cortex
(21)
.
Finally, we integrate these analyses to provide a map of
functional interactions (
Fig. 7.10
). In this case, we are not
interested in significance; rather, we are interested in patterns of
interactions including subthreshold values. For these reasons, we
present a scaled correlation map across ensembles of neurons in
frontal cortex.
We note that, in this data set, functional interactions in motor
cortex common at the level of cross-correlation and JPSTH, but
rare in rate or predictive interactions. On the contrary, in dmPFC,
cross-correlation and JPSTH interactions are rare, but rate or
predictive interactions are more common. Between dmPFC and
motor cortex, particular combinations of neurons seemed to
have robust functional correlations. These results demonstrate,
Fig. 7.10. Integrating functional interactions between measures. (A) Interactions within
motor cortex; (B) dmPFC, and between dmPFC and motor cortex using techniques pre-
sented in this paper. For comparison, all interactions are normalized within a class of
interactions, and presented as absolute value.
