Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
4336036 | Journal of Neuroscience Methods | 2008 | 7 Pages |
We present a statistical approach to the identification of correlated activity in multineuron spike data, based on the value of the correlation determinant. This approach is not compromised by the lack of independence often encountered in this kind of data. We illustrate our method by applying it both to simulated data and to data recorded from neurons in a forebrain region (intermediate medial mesopallium, IMM) of the behaving domestic chick and simultaneously from the corresponding contralateral region. There is no direct anatomical connection between the two sites, and the validity of this technique is strongly supported by the observation that when the test indicates significantly correlated activity for neurons within either hemisphere, this correlation is greatly reduced, and ultimately obliterated, by serial incorporation of activity from neurons in the opposite hemisphere. Since the value of individual correlation coefficients allied to the Bonferroni correction is often used as a diagnostic tool, we also present comparisons of that approach with our correlation determinant approach.