Critical to vision research is the generation of visual displays with precise control over stimulus metrics. Generating stimuli often requires adapting commercial software or developing specialized software for specific research applications. In order to facilitate this process, we give here an overview that allows nonexpert users to generate and customize stimuli for vision research. We rust give a review of relevant hardware and software considerations, to allow the selection of display hardware, operating system, programming language, and graphics packages most appropriate for specific research applications. We then describe the framework of a generic computer program that can be adapted for use with a broad range of experimental applications. Stimuli are generated in the context of trial events, allowing the display of text messages, the monitoring of subject responses and reaction times, and the inclusion of contingency algorithms. This approach allows direct control and management of computer-generated visual stimuli while utilizing the full capabilities of modern hardware and software systems. The flowchart and source code for the stimulus-generating program may be downloaded from www.psychonotnic.org/archive.We give here an overview of practical considerations for generating customized graphics for vision research. We include a brief overview of hardware considerations, followed by a discussion of programming strategies that allow specification of stimuli while utilizing memory and processing capacities of modem graphics hardware. We also include a discussion of suitable application programming interfaces (APIs) in the Windows operating system. We make recommendations for software implementations, but we also provide information needed for selecting an approach suitable for specific needs. Finally, we provide specific code that can be used as a framework for generating visual stimuli and controlling trial events. Programming examples address graphics control, animation techniques, color palettes, and timing techniques. We provide programming details to allow conversion of DOS-based VGA graphics to Win32 programs, or for generating new programs for a broad range of research applications.Commercially available software for application in vision research (see a representative list in Table 1) typically provides a library of common stimuli that may be modified within a framework. Such software does not provide the flexibility to generate unique stimuli required for novel and evolving research areas. Therefore, customized programming for generating stimuli within the context of specific research applications is needed. Programming examples provided here are designed for researchers with basic experience in computer programming. Programming examples describe the generation of simple stimuli, whereas the general approach may be used to generate complex and dynamic stimuli.

Across the visual field, progressive differences exist in neural processing as well as perceptual abilities. Expansion of stimulus scale across eccentricity compensates for some basic visual capacities, but not for high-order functions. It was hypothesized that as with many higher-order functions, perceptual grouping ability should decline across eccentricity. To test this prediction, psychophysical measurements of grouping were made across eccentricity. Participants indicated the dominant grouping of dot grids in which grouping was based upon luminance, motion, orientation, or proximity. Across trials, the organization of stimuli was systematically decreased until perceived grouping became ambiguous. For all stimulus features, grouping ability remained relatively stable until 40°, beyond which thresholds significantly elevated. The pattern of change across eccentricity varied across stimulus feature, in which stimulus scale, dot size, or stimulus size interacted with eccentricity effects. These results demonstrate that perceptual grouping of such stimuli is not reliant upon foveal viewing, and suggest that selection of dominant grouping patterns from ambiguous displays operates similarly across much of the visual field.

Early visual processing in rats is mediated by several pre-cortical pathways as well as multiple retinal ganglion cell types that vary in response characteristics. Discrete processing is thereby optimized for select ranges of stimulus parameters. In order to explore variation in response characteristics at a perceptual level, visual detection in rats was measured across a range of contrasts, spatial frequencies, and durations. Rats responded to the onset of Gabor patches. Onset time occurred after a random delay, and reaction time (RT) frequency distribution served to index target visibility. It was found that lower spatial frequency produced shorter RTs, as well as increased RT equivalent of contrast gain. Brief stimulus presentation reduced target visibility, slowed RTs, and reduced contrast gain at higher spatial frequencies. However, brief stimuli shortened RTs at low contrasts and low spatial frequencies, suggesting transient stimuli are more efficiently processed under these conditions. Collectively, perceptual characteristics appear to reflect distinctions in neural responses at early stages of processing. The RT characteristics found here may thereby reflect the contribution of multiple channels, and suggest a progressive shift in relative involvement across parameter levels.

No abstract

Mechanisms underlying perceptual grouping serve to bind stimulus components that are contained within grouped patterns. In order to examine the time course of grouping development, grids of spatially isolated dots were followed by pattern masks across a range of SOA. Subjects indicated the predominant perceived grouping of the dot patterns. Masks either spatially superimposed target elements (element mask), or superimposed elements as well as paths among elements (connection mask). Element masks thereby disrupted processing of target elements, while connection masks additionally disrupted representations in regions among elements. It was found that element masks disrupted grouping 12ms after target offset, after which masks had no effect. Connection masks disrupted grouping up to 47ms following target offset. Results suggest grouping mechanisms access the afferent signal for a brief period early in processing, after which binding formation proceeds for an addition 35ms. Shortening connection mask duration to 12ms enhanced performance during a brief temporal window within the interference period. For each set of conditions, target elements were visible during the time frame in which stimulus patterns could not be perceptually grouped. Full-field checkerboard masks degraded discrimination similarly as connection masks, although were more effective in disrupting discrimination with an SOA of 24 and 36ms. Degrading stimulus organization progressively extended the time scale for each masking effect. For the grouping of low-level stimulus features tested here, results support a model in which afferent signals are accessed early, followed by progressive binding among grouped elements. Effect of shortening connection masks may reflect incomplete disruption of target processing, or possibly re-entry of stimulus representations by feedback from higher processing areas.