Biomedical Engineering Reference
In-Depth Information
in principle, how functional interactions can be combined to elu-
cidate patterns of interaction among cortical networks.
Thus, studying different kinds of functional interactions
across an ensemble of simultaneously recorded neurons in multi-
electrode experiments affords novel insights into how neurons
work together to encode information, and provides evidence that
neurons can work together in diverse ways in service of behavior.
12. Notes
Here, we have described seven distinct approaches to investigat-
ing functional interactions between neurons: Functional grouping
by PCA, cross-correlation, joint-peristimulus time histograms,
rate correlations, trial-by-trial rate correlations, predictive inter-
actions, and network interactions. We have described how to per-
form each analysis, limitations and applications of each analysis,
and demonstrated how such analysis might work in an ensemble
of 21 simultaneously recorded neurons from rodent dmPFC and
motor cortex. By integrating and comparing functional interac-
tions, we are able to gain insight on how relationships among
populations of neurons contribute to the encoding of behavior.
While we have restricted our analysis of functional interac-
tions to the methods discussed here, many investigators have
developed an arsenal of tools with which to assail the prob-
lem of neuronal interactions. These include peer-prediction, in
which pattern recognition is used to predict the spikes of another
neuron
(39)
; information theory, in which a 'lookup table' is
constructed to predict the information in one spike train from
another
(37)
, detection of synchrony
(40)
, coherence (both in
spike and field), in which frequency relationships between field
potential and spikes from different channels are used to make
inferences about local and distant processing
(41)
,aswellasa
host more under current development. Each of these techniques
has strengths and pitfalls, and may be able to be readily applied
to our data. However, though we have limited our discussion
to several readily applicable and relatively simple techniques that
demonstrate the principles of considering functional interactions
between neurons, we recognize that future analyses using the
techniques above or still being developed may further refine or
expand our interpretations of how neurons interact.
What is the significance of the functional interactions
described here? Averbeck and colleagues have examined corre-
lated variability between neurons
(15-17, 38)
. They mention two
ideas when considering neuronal correlations:
Signal
correlations,
i.e., correlations that relate to an outside, measured variable;
and
noise
correlations, i.e., correlations that do not relate to any
