Figure 3. Predicting the responses of a simple cell based on an RF model of a 2D symmetric Gabor model. A, A 2D-oriented Gabor function: a sinusoidal plane wave weighted by a Gaussian envelope in two different phases: θ = 0 (sine wave) or θ = π/2 (cosine wave) shown at the top and bottom panels that generate Gabor filters with odd and even symmetry, respectively. B, Constructing spatiotemporally oriented impulse responses from gratings stimuli drifting against a Gabor RF. Top, Examples of stimuli with four different orientations. Bottom, Left, An example of a Gabor RF with θ = π/2. Bottom, Right, Four examples where each box depicts the overlap of the grating stimulus crossing the RF in a phase that yields the maximum response, calculated as the inner product between the stimulus and the RF. C, A predicted response tuning matrix computed as the inner product between the Gabor RF model shown in A, and stimuli of square-wave gratings with various orientations (1º interval) and seven SFs. Each pixel represents the inner product between the RF and the stimulus (maximized across phase, see Materials and Methods). The row marked with a blue arrow on the right denotes the preferred SF and was taken as a reference for comparing other SFs. Right, Orientation tuning curves at various SFs, color coded according to the arrows shown next to the predicted tuning matrix. Top and bottom panels correspond to RF with θ = 0 and θ = π/2, respectively. a.u., Arbitrary units. D, Comparison of orientation tuning parameters between various SFs, based on the predicted response shown in C. Blue and red lines depict the parameters calculated based on a Gabor RF model with θ = 0 and θ = π/2, respectively. Shown are the ΔF/F to the Pref, the ΔF/F to the Orth, OSI, 1 − CirVar, HWHH, and the preferred orientation.