Biomedical Engineering Reference
In-Depth Information
or relate LFP spectral power in a particular frequency to behavior. Regression techniques are also
used to relate observed neural activity to continuous behavioral information (i.e., hand trajectory).
Ultimately, hypothesis testing is required to validate inference-based approaches where significance
is used to distinguish between behavioral conditions and neural activity.
2.7 METhodS oF KINEMaTIC aNd dyNaMIC
REPRESENTaTIoN
The firing of neurons in the primary motor (M1), premotor (PM), parietal (P), somatosensory
(S) cortices, basal ganglia (BG), and cerebellum (CB) contains the representation of many motor
and nonmotor signals. Through single-cell and neural population electrophysiological experiments,
neuronal firing has been related to joint angles [ 40 ], muscle activation [ 41-43 ], hand trajectory
(path) [ 44-46 ], velocity [ 47 , 48 ], duration [ 49 , 50 ], as shown in Figure 2.6 . From a computational
perspective, the challenge has been one of understanding how information flows in different sensory
and motor networks, making contributions to each of the parameters in a time-varying manner.
Several hypotheses have been proposed about how kinematic and dynamical variables are repre-
sented in the nervous system. Up to this point, we have discussed that information processing in the
brain revolves around modulation (variation) in neuronal firing. This variation could be locked to
some external stimulus or even the firing of another neuron. The three leading hypotheses of neuron
information processing include rate coding, correlation coding, and temporal coding. A detailed
discussion of the analysis approaches for each hypothesis will be given next.
FIgURE 2.6: Parameters of cell signaling in the CNS [ 51 ].
 
Search WWH ::




Custom Search