| Perceptual performance is a function of signal contrast, external noise potency, template overlap, and the observer state. Many observer models (e.g., the Perceptual Template Model (Lu & Dosher, 1998)), were constructed for cases where a stimulus activates only one perceptual template. Here, we developed a model to handle cases when a stimulus activates overlapping perceptual templates, so that the model can predict the perceptual performance in both contrast and feature domains.;To validate the model, contrast threshold versus external noise contrast (TvC) functions were measured first in an orientation identification task at fovea for four orientation differences (+/-3°, +/-6°, +/-15°, and +/-45°) across external noises, using the Method of Constant Stimuli (MCS) for three observers. Multiple- TvCs (at 65, 75, and 85% performance levels) were obtained for each orientation separation condition. The model provided excellent fits to the data with the least number of parameters. The model successfully described identification task performance with signal contrast, external noise contrast, and template overlaps.;In principle, the model also supports the estimation of feature thresholds. Unfortunately, measuring a multiple-TvCs in several external noise levels requires large investment in data collection. Therefore, more efficient and quicker method to estimate TvC was imperative for one to study other perceptual processes such as perceptual learning, using the external noise paradigm. As a counterpart of the quick-TvC (Lesmes, Jeon, Lu & Dosher, 2006) developed in the contrast domain, quick-fTvC (QfTvC) is developed to measure thresholds in the feature domain more effectively.;First, the regularities of TvC shown in the contrast domain were validated in the feature domain. Then, the QfTvC was validated with simulations and psychophysical experiments. Simulations showed that 300 trials were enough to estimate TvC functions at widely separated performance criteria with good accuracy and precision. Psychophysical validation showed very good agreement between the TvC estimates from the QfTvC and from the MCS, although the QfTvC used only 10% of trials used in the MCS (300 vs 3360 trials). The QfTvC may contribute significantly practical value for applying the external noise method to study mechanisms of observer-state changes and special population. |
