Skip to main content
. Author manuscript; available in PMC: 2015 Aug 1.
Published in final edited form as: J Neural Eng. 2014 Jun 12;11(4):046007. doi: 10.1088/1741-2560/11/4/046007

Fig. 5. Reliability of LFP-based offline decoding is similar to spike-based offline performance.

Fig. 5

A. Directional decoding using a classifier recalibrated in each research session. Error bars indicate 95% confidence intervals obtained from bootstrapping the five-fold cross validation results. Dashed lines show chance level B. Offline decoding performance using the same classifier calibrated on the first session. C. Number of significantly modulated features co-vary between spikes and LFPs. D. Summary performance of the five classifiers tested over the study period reveals similar performance for spikes and LFPs. E. Offline decoding performance using the same decoder decreases by 60% over time periods of ∼2 months but remains significantly different from chance level (Methods) both in spikes and LFPs. Larger, filled symbols indicate statistically different performance from chance level. Fraction change indicates the proportional change in decoding performance respect to the calibration session.