Does the sluggishness of colour perception reflect a limitation in the maximum (slew) rate at which chromatic pathways respond to changes in colour?

At higher temporal frequencies, the appearance of flickering lights is dominated by a steady average, with some residual disturbance at the flicker frequency. The average appearances of flickering lights that turn rapidly off and slowly on and those that turn slowly on and rapidly off can be very different even when their mean colours and intensities are the same. (We refer to these waveforms as 'sawtooth' stimuli.) These differences, which are unexpected, are important because they can provide unique insights into the underlying processing of visual signals that encode colour. Our initial hypothesis is that the differences are due to chromatic mechanisms being better able to follow slow colour changes than fast ones. As a result, their mean outputs become skewed in the direction of the slow change. For example, a yellow light that repeatedly changes slowly to red but then rapidly to green looks on average more red than a light that changes rapidly to red but slowly to green, which looks more green. This imbalance is inconsistent with many current models of chromatic processing, but can be explained by our model. Specifically, we propose that the rate at which colour mechanisms can signal changes in colour is limited to a particular level. If the rate of change exceeds that level, the output varies only at the limiting rate of change. We call this the 'slew-rate' model. Although the idea of a slew-rate limit is familiar in electronics (for example in the specification of operational amplifiers), it is as far as we know novel in the field of vision research. We propose a series of experiments that are designed to test and develop the slew-rate model. Measurements of chromatic detection and chromatic discrimination will be made to characterize the effects of changing the slopes of the sawtooth waveforms as a function of various stimulus parameters. At each stage, these results will be compared with predictions generated by computer simulations of the slew-rate model. Our goal is to critically evaluate the slew rate model. To what extent can the model account for the data? Can simple modifications to the model better account for the data? Can other types of waveforms be used to critically test the model? Can computer simulations suggest other experiments? In addition to causing changes in the mean colour of sawtooth waveforms, a limiting slew-rate should also make us less sensitive to flicker. Thus, the well-known sluggishness of colour perception may be due to the slew-rate limit rather than the more conventional explanation that colour signals are averaged over time. We will also test this prediction by comparing flicker sensitivity measurements with the sawtooth measurements. Testing the slew-rate model per se is not necessarily the most critical feature of this proposal. Equally critical is the fact that we are generating a new dataset using unconventional stimuli and an unconventional model and approach. This new dataset will be made available to other researchers, and can be used by others in the field as well as ourselves to generate new models of how colour vision works. The primary goal of this proposal is to understand more about how chromatic signals are processed in the human brain and in particular why chromatic signals are processed much more slowly than achromatic ones. Despite the many recent advances in related fields, details about how colour processing in humans work remains relatively little understood.

In general, we are relatively less sensitive to high temporal-frequency chromatic flicker than we are to high temporal-frequency achromatic flicker. The relative loss of chromatic sensitivity is usually taken as evidence that the chromatic response is somehow 'slower' than the achromatic one. But in what way is colour processing 'slow'? In the linear-systems view, the additional slowness of chromatic signals is attributed to chromatic mechanisms having longer integration times, either because of additional filtering stages and/or stages with longer time constants. In this proposal, wWe should like to take advantage of a novel chromatic phenomenon that is inconsistent with this type of model: The new phenomenon is that the mean colour appearance of rapidly-on and slowly-off sawtooth flicker |\|\|\|\|\ can be markedly different from that of slowly-on and rapidly-off sawtooth flicker /|/|/|/|/|, even though the waveforms have the same mean luminance and chromaticity. Although inconsistent with conventional models, the phenomenon is consistent with a simple model in which the chromatic mechanisms are assumed to be limited by the rate at which they can signal changes in colour (slew-rate limitation). Thus, the chromatic mechanisms are better able to track the slowly changing part of the sawtooth than its quickly changing part, so that their mean output will always be skewed in the direction of the slow change. In this project, we propose to develop and test 'slew-rate-limit' models. We will use objective two-alternative forced choice (2AFC) measurement techniques to systematically characterize the phenomenon as a function of various stimulus parameters. In parallel with these measurements, we will generate computer simulations of the effects of a limiting slew-rate in order to produce testable predictions. The model will be adapted or revised as necessary. This project will, we believe, shed light on fundamental aspects of human colour vision.

The research will directly benefit vision scientists, sensory physiologists, and cognitive neuroscientists working on chromatic processing in the human and primate visual system in several ways. First, it will benefit vision scientists and psychophysicists by providing a better understanding of how chromatic signals are processed and limited by postreceptoral processing and what makes them slower than achromatic signals. Second, and in parallel, it will benefit retinal physiologists by providing them with behaviourally-measured baselines and new techniques and stimuli to evaluate and guide their single-cell recording experiments. The approach of viewing their results as slew-rate limited may lead to exciting and fruitful new avenues of research. Third, it will provide cognitive neuroscientists with new information about the types and nature of chromatic signals that are likely to be encountered in the human cortex. Many other scientists, who are less directly involved in colour vision research, including some geneticists, anatomists, brain-imaging specialists, and electrophysiologists will also potentially benefit from this research. At the more applied level, better understanding of chromatic processing will be useful in machine-computer interfaces. Knowing the limits of chromatic processing will help to define the capabilities required of such interface systems. For example, if the colour mechanisms are limited by a particular slew rate, there is no need for the interface device or other display device to exceed that rate. Such interfaces and displays include videos, films, mobile phones and other dynamic presentation devices. The work may also provide important insights into efficient algorithms for extracting chromatic information in artificial vision systems. Given the wide range of interest in colour vision across many disciplines it will be important that our results and models are widely disseminated. We will publish our work in open source journals and present the data at international conferences. As part of this project, we also propose to make our data available at the Colour and Vision Research Laboratory (CVRL) website run by the PI at http://www.cvrl.org This resource is well-known and widely used in colour research both by academics and in industry. We propose to develop this resource further to provide general information on vision.