Published in Computational and Systems Neuroscience (CoSyNe), Feb 2017.
Many visually-driven activities require observers to integrate information over time. Traditional models suggest that the signal-to-noise ratio (SNR) of integrated sensory information should improve in proportion to the square root of integration time. However, in experimentally controlled tasks, observer performance grows at a substantially slower rate, often saturating over relatively short time scales. This failure has been ascribed to multiple causes, including rapid sensory adaptation, faulty accumulation of available information, and incorporation of task-related incentives for rapid decision-making. But temporal integration may be limited by a more elementary factor: Temporal correlation of noise (Osborne, Bialek & Lisberger, 2004). We examined responses of orientation-selective neurons in the primary visual cortex (V1) of two macaque monkeys performing an orientation discrimination task. The signal-to-noise ratio (SNR) of temporally integrated responses increased for durations up to approximately 250 ms, but saturated for longer durations. This was true even when cells exhibited little or no adaptation in their response levels. Analysis of another data set revealed similar effects in extrastriate area MT. We show that the saturation of SNR is consistent with a previously proposed response model in which spikes arise from a Poisson process whose stimulus-dependent rate is modulated by slow, stimulus-independent fluctuations in gain (Goris, Movshon & Simoncelli, 2014). Specifically, the response variability arising from the Poisson process is reduced by temporal integration, but the temporally correlated variability due to gain fluctuations is not. A model with slow additive fluctuations does not exhibit these behaviors. Our analysis suggests that slow gain fluctuations impose a fundamental limit on the benefits of temporal integration throughout the visual cortex.