Accurate saccadic programming in natural visual scenes requires a signal designating which of the many potential targets is to be the goal of the saccade. Is this signal controlled by the allocation of perceptual attention, or do saccades have their own independent selective filter? We found evidence for the involvement of perceptual attention, namely: (1) summoning perceptual attention to a target also facilitated saccades; (2) perceptual identification was better at the saccadic goal than elsewhere; and (3) attempts to dissociate the locus of attention from the saccadic goal were unsuccessful, i.e. it was not possible to prepare to look quickly and accurately at one target while at the same time making highly accurate perceptual judgements about targets elsewhere. We also studied the trade-off between saccadic and perceptual performance by means of a novel application of the "attentional operating characteristic" (AOC) to oculomotor performance. This analysis revealed that some attention could be diverted from the saccadic goal with virtually no cost to either saccadic latency or accuracy, showing that there is a ceiling on the attentional demands of saccades. The links we discovered between saccades and attention can be explained by a model in which perceptual attention determines the endpoint of the saccade, while a separate trigger signal initiates the saccade in response to transient changes in the attentional locus. The model will be discussed in the context of current neurophysiological work on saccadic control.

Subjects viewed sequentially presented lists of 3-6 words, which were followed by a recognition probe. Memory retrieval speed (dynamics) and strength were measured in an interruption speedaccuracy trade-off (SAT) procedure and a collateral reaction time (RT) procedure. In SAT, item strengths depended on serial position, but only two retrieval speeds were observed: a fast rate for the last item in the study list (a case of immediate repetition between study and test) and a slow rate for all other items that was independent of serial position and set size. Serial-positiondependent strengths and set-size-dependent criterion shifts accounted for standard RT patterns that have been taken as evidence for serial scanning in short-term memory. Summary of ExperimentsWe examined retrieval in immediate memory for short lists, using speed-accuracy trade-off (SAT) and comparable reaction time (RT) paradigms. We replicated MonselFs (1978) demonstration of strong recency or serial position effects on RT, which average to produce linear, parallel set size functions for both positive and negative trials. In SAT, memory strength or probability (indexed by asymptotic accuracy) and retrieval speed (indexed by rate and intercept) were separately estimated for full retrieval functions. An SAT experiment showed that the serial position effects in RT are paralleled by analogous effects on asymptotic memory strength. The dynamics of retrieval (rate and intercept) were equal for all serial positions except the most recent and were independent of set size. There was a large speeding in retrieval when the test item was the same as the last list member (immediate repetition), which is a replication of a similar finding by Wickelgren, Corbett, and Dosher (1980) in long lists. Thus observed SAT rate differences between set sizes when serial position data were pooled reflected the relative proportion of immediate repetitions in different set sizes. Set size per se had no effect on retrieval speed, although different serial position mixes resulted in different asymptotic accuracy. The immediate repetition effect did not depend on a physical match between the last list element and the test. All results were replicated in a second SAT experiment, in which we presented list items in lowercase letters and test items in uppercase letters. In that experiment we also examined the effect of recency on negative trials. Recent negatives (items presented in the immediately previous memory list) yielded higher false alarms than did distant negatives, especially early in retrieval. Serial-scanning

To investigate the nature of plasticity in the adult visual system, perceptual learning was measured in a peripheral orientation discrimination task with systematically varying amounts of external (environmental) noise. The signal contrasts required to achieve threshold were reduced by a factor or two or more after training at all levels of external noise. The strong quantitative regularities revealed by this novel paradigm ruled out changes in multiplicative internal noise, changes in transducer nonlinearites, and simple attentional tradeoffs. Instead, the regularities specify the mechanisms of perceptual learning at the behavioral level as a combination of external noise exclusion and stimulus enhancement via additive internal noise reduction. The findings also constrain the neural architecture of perceptual learning. Plasticity in the weights between basic visual channels and decision is sufficient to account for perceptual learning without requiring the retuning of visual mechanisms.

We developed and tested a powerful method for identifying and characterizing the effect of attention on performance in visual tasks as due to signal enhancement, distractor exclusion, or internal noise suppression. Based on a noisy Perceptual Template Model (PTM) of a human observer, the method adds increasing amounts of external noise (white gaussian random noise) to the visual stimulus and observes the effect on performance of a perceptual task for attended and unattended stimuli. The three mechanisms of attention yield three "signature" patterns of performance. The general framework for characterizing the mechanisms of attention is used here to investigate the attentional mechanisms in a concurrent location-cued orientation discrimination task. Test stimuli--Gabor patches tilted slightly to the right or left--always appeared on both the left and the right of fixation, and varied independently. Observers were cued on each trial to attend to the left, the right, or evenly to both stimuli, and decide the direction of tilt of both test stimuli. For eight levels of added external noise and three attention conditions (attended, unattended, and equal), subjects' contrast threshold levels were determined. At low levels of external noise, attention affected threshold contrast: threshold contrasts for non-attended stimuli were systematically higher than for equal attention stimuli, which were, in turn, higher than for attended stimuli. Specifically, when the rms contrast of the external noise is below 10%, there is a consistent 17% elevation of contrast threshold from attended to unattended condition across all three subjects. For higher levels of external noise, attention conditions did not affect threshold contrast values at all. These strong results are characteristic of a signal enhancement, or equivalently, an internal additive noise reduction mechanism of attention.

The mechanisms of perceptual learning are analyzed theoretically, probed in an orientation-discrimination experiment involving a novel nonstationary context manipulation, and instantiated in a detailed computational model. Two hypotheses are examined: modification of early cortical representations versus task-specific selective reweighting. Representation modification seems neither functionally necessary nor implied by the available psychophysical and physiological evidence. Computer simulations and mathematical analyses demonstrate the functional and empirical adequacy of selective reweighting as a perceptual learning mechanism. The stimulus images are processed by standard orientation- and frequency-tuned representational units, divisively normalized. Learning occurs only in the "read-out" connections to a decision unit; the stimulus representations never change. An incremental Hebbian rule tracks the task-dependent predictive value of each unit, thereby improving the signal-to-noise ratio of their weighted combination. Each abrupt change in the environmental statistics induces a switch cost in the learning curves as the system temporarily works with suboptimal weights.

External noise methods and observer models have been widely used to characterize the intrinsic perceptual limitations of human observers and changes of the perceptual limitations associated with cognitive, developmental, and disease processes by highlighting the variance of internal representations. The authors conducted a comprehensive review of the 5 most prominent observer models through the development of a common formalism. They derived new predictions of the models for a common set of behavioral tests that were compared with the data in the literature and a new experiment. The comparison between the model predictions and the empirical data resulted in very strong constraints on the observer models. The perceptual template model provided the best account of all the empirical data in the visual domain. The choice of the observer model has significant implications for the interpretation of data from other external noise paradigms, as well as studies using external noise to assay changes of perceptual limitations associated with observer states. The empirical and theoretical development suggests possible parallel developments in other sensory modalities and studies of high-level cognitive processes.Keywords: signal detection theory, internal representation, observer model, perceptual template, internal noise Human decisions are based on internal representations of information. Understanding how stimuli are represented internally is one of the classic problems in psychology. This article examines how modifying an external stimulus with external noise can provide insight into how the stimulus is processed by the human observer. We conducted a systematic and comprehensive review of the external noise paradigms and observer models widely used in characterizing the internal response properties of human observers. The observer approach builds on the broadly applicable framework of signal detection theory (SDT) by elaborating the relationships between external stimuli and the internal response distributions that form the basis for decision. The empirical tests introduce external noise-either masking noise or variation in the relevant stimulus dimension-to provide a reference for characterizing and quantifying the limiting factors in perceptual sensitivity. The current review, analysis, and empirical test focus on visual perception. Some of the model properties, especially the empirical findings, may be modality specific. Still, this framework and the findings could serve as an example for parallel development of the empirical methods and theoretical models in other sensory modalities. The model development and testing also have major implications for applications of the external noise paradigms in understanding the mechanisms underlying changes of perceptual sensitivity in different cognitive, disease, and/or developmental states. Internal Response DistributionsSDT provides a general framework for analyzing human decision making in perceptual and cognitive tasks (Green & Swets, 1966;Macmillan & Creelman, 1991). In a ...

Based on eye-fixation patterns, strategies for multiattribute binary choice were classified as holistic (within an alternative) or dimensional (within an attribute across alternatives). In a task environment hospitable to both strategies, dimensional processing predominated. Even for alternatives like simple gambles, which require holistic computations, dimensional strategies were used as often as holistic ones. The dimensional strategies were augmented by two procedures that simplify the computations. These simplification procedures reduce cognitive effort at the cost of a relatively small increase in errors. However, for about half the subjects the use of these simplification procedures led to systematic violations of expected utility theory on certain choices. Both the preference for dimensional over holistic strategies and the adoption of simplifying procedures are compatible with the desire to reduce cognitive effort. We propose that strategies are selected to minimize the joint cost of errors and effort.

Abstract-Precue