Purpose: We measured reaction times (RT) for inferred rod and cone ON and OFF pathways. A neural model for RT was developed based on rod and cone impulse response functions.

Method: A rod or cone (L+M+S) isolating stimulus (2° center, 13°surround, 7.5° eccentricity) was presented using a 2-channel, 4-primary colorimetric system. Retinal illuminance spanned a 5-log unit range (0.002-200 Td). RTs were measured at 5 contrasts, using 1-sec rapid-on or rapid-off ramp waveforms. The measured RTs were modeled based on rod and cone impulse response functions derived from published psychophysical temporal contrast sensitivity data, and a decision-making process that depended on the integrated neural signals.

Results: RTs decreased with increasing contrast and retinal illuminance. RTs to incremental and decremental cone-isolating stimuli were similar at all light levels. At low light levels (≤ 0.2 Td), rod responses for decrements were faster than for increments. RTs for rod and cone stimuli were comparable at 2 Td; cones responded faster than rods at light levels ≥20 Td. Our model described the data using a single value across conditions specifying the irreducible minimum reaction time, and one free parameter for each increment/decrement/light level condition.

Conclusion: The reaction times of cone ON and OFF pathways were similar. Rod OFF pathway responses were faster than ON pathway responses at low light levels. Cones were faster than rods only above high mesopic light levels. Our impulse-response-function based model captures all attributes of the RT data with a limited number of free parameters.