We first mapped response fields of neurons in LIP in both tasks by placing the target at any one of the eight peripheral targets. After determining the response field, we presented the target at the location in the response field of the LIP neuron under study (preferred direction) or the opposite location (null direction). Preferred and null target locations were interleaved trial-by-trial in equal proportions. Neural recordings were made using multiple-electrode

microdrives (Double MT, Alpha Omega, Israel). Spiking and LFP activity were recorded with glass-coated tungsten electrodes (Alpha Omega, Israel) with impedance 0.7–1.4 MOhm measured at 1 kHz (Bak Electronics, MD). Neural signals were amplified (×10,000; Ruxolitinib TDT Electronics, Alachua, FL), digitized at 20 kHz (National Instruments), and continuously streamed to disk during the experiment (custom C and Matlab code). Broadband neural activity was preprocessed to obtain single-unit

spike times and LFP activity. Recordings in area LIP and V3d were acquired with respect to a reference placed at the cortical surface on the lateral bank of the intraparietal sulcus. Recordings in PRR were acquired with respect to a reference placed at the cortical surface on the medial bank of the intraparietal sulcus. See also, Supplemental Experimental Procedures. To analyze the relationship between RTs and LFP power at each time and frequency, we subtracted LFP power before movements in trials with the slowest 33% of RTs from GW786034 LFP power before movements in trials with the fastest 33% of RTs and computed a z-score using 1,000 random permutations (Maris et al., 2007). By fixing the proportions of trials across sessions, we were able to effectively control the degree that RT differed between fast and slow trial groups. The z-score was

approximated to be normally distributed with mean 0 and variance 1, and values with an absolute value greater than 1.96 were taken to be significant with probability p < 0.05. Similarly, to examine the spatial selectivity of LFP power at each time and frequency, we subtracted LFP power before movements in the null direction from LFP power before movements in the preferred direction and computed a z-score using 1,000 random permutations. We confirmed that the null distribution unless of permuted power differences satisfied the normal approximation (Kolmogorov-Smirnow test, p > 0.05). Correlations between SRT and RRT were calculated using Pearson’s correlation coefficient. Similar results were obtained using Spearman’s Rank correlation coefficient (data not shown). To examine the relationship between SRT and RRT while controlling for beta-band power, we estimated RT correlations across groups of trials when beta-band power was held constant and compared the results with RT correlations across groups of trials when beta-band power varied.