Domain size was calculated by averaging each domain’s long and short axes measured from the t-map using ImageJ (National Institutes of Health). Classical single-cell extracellular recordings (Hubel and Wiesel, 1968) were performed in three hemispheres of three anesthetized macaques. A recording chamber and a silicon hat (Arieli et al., 2002) were implanted following the initial optical imaging session. Electrode penetrations
were made at specific V4 regions (either at the center of a direction-preferring domain or at a Raf inhibitor region far away from direction-preferring domains) guided by the cortical blood vessel patterns. Tungsten microelectrodes (impedance 1–4 MΩ at 1 kHz, FHC) were lowered into
the cortex using a hydraulic microdrive (MO-97A, Narishige). Neural activity was amplified at 10 kHz (Model 1800, A-M Systems) and digitized at a sampling rate of 20 kHz (Power 1401, CED). Single-cell activity was isolated and sorted online (Spike2, CED). Once a single cell was isolated, its classical receptive field was plotted using a manually controlled bar and/or grating stimulus. Sine-wave or square-wave gratings drifting in eight different directions were then displayed within the cell’s receptive field to measure orientation selleck and direction preferences. The spatial and temporal frequencies of the gratings were adjusted to best drive the cell. Each Adenosine stimulus presentation lasted 1 s and was followed by a 1 s ISI. In some experiments, we also used a 0.5 s stimulus presentation with a 1.5 s ISI. Usually, 10–25 repeats were collected for each stimulus condition. Neuronal responses to each direction were calculated by averaging the spike numbers during the stimulus presentation, shifted by a delay of 100 ms. The tuning functions were then fitted with a modified von Mises curve (Mardia, 1972) which fits well with both unimodal and bimodal distributions: y=a+b1∗ec1∗cos(x−d1)+b2∗ec2∗cos(x−d2)y=a+b1∗ec1∗cos(x−d1)+b2∗ec2∗cos(x−d2), in which x is the direction tested; y is the corresponding firing rate and is
a function of x; a is the baseline offset; and (b1, b2), (c1, c2), and (d1, d2) determine the amplitude, shape, and position of the tuning curve, respectively. Fitting parameters were obtained with a least-square nonlinear regression method (nlinfit in Matlab, Mathworks). Goodness of fit (R2) values were >0.7 for all units (n = 63) and were >0.9 for 52 out of 63 units. Each neuron’s direction-of-motion selectivity was determined using a DI based on a fitted response profile: DI = 1 − Rn/ Rp, in which Rp is the response to the preferred direction (direction that generated the maximum response) and Rn is the response to the null direction (direction that opposite to the preferred direction). DI values range from 0 to 1, with 1 being the maximum directional selectivity.
No related posts.