Attentional demands following perceptual skill training.

Practicing simple visual tasks induces substantial improvement. We investigated whether increased efficiency is accompanied by automaticity and immunity to across-task interference. We found that although practice speeds orientation feature detection, it does not abolish susceptibility to interference from introduction of concurrent central-letter identification, which takes priority. Yet following training with each task observers successfully managed to perform the tasks concurrently. The effectiveness of separate training implies that the role of improved intertask coordination in achieving concurrent performance was minor. Indeed, even when initial training was concurrent, improvement on the two tasks was sequential, and the higher-priority (central) task was learned first. However, automatic processing was not accomplished either, because increasing the difficulty of the higher-priority task interfered with performance of both tasks. What appears to be orchestrated posttraining performance is actually mainly an emergent property of speeded initial processes rather than either eliminated bottlenecks or improved central executive management.

Forming classes by stimulus frequency: behavior and theory.

Visual classification is the way we relate to different images in our environment as if they were the same, while relating differently to other collections of stimuli (e.g., human vs. animal faces). It is still not clear, however, how the brain forms such classes, especially when introduced with new or changing environments. To isolate a perception-based mechanism underlying class representation, we studied unsupervised classification of an incoming stream of simple images. Classification patterns were clearly affected by stimulus frequency distribution, although subjects were unaware of this distribution. There was a common bias to locate class centers near the most frequent stimuli and their boundaries near the least frequent stimuli. Responses were also faster for more frequent stimuli. Using a minimal, biologically based neural-network model, we demonstrate that a simple, self-organizing representation mechanism based on overlapping tuning curves and slow Hebbian learning suffices to ensure classification. Combined behavioral and theoretical results predict large tuning overlap, implicating posterior infero-temporal cortex as a possible site of classification.

The spread of attention and learning in feature search: effects of target distribution and task difficulty.

We examined the roles of two determinants of spatial attention in governing the spread of perceptual learning, namely, stimulus location distribution and task difficulty. Subjects were trained on detection of a target element with an odd orientation imbedded in an array of light bars with otherwise uniform orientation. To assess the effects of target distribution on attention and learning, target positions were distributed so that attention was allocated not only to the target positions themselves, but also to intermediate positions where the target was not presented. Target detection performance substantially improved and improvement spread to match the induced window of spatial attention rather than only the actual target locations. To assess the effect of task difficulty on the spread of attention and learning, the target-distractor orientation difference and the time interval available for processing were manipulated. In addition, we compared performance of subjects with more versus with less detection difficulty. A consistent pattern emerged: When the task becomes more difficult, the window of attention shrinks, and learning becomes more localized. We conclude that task-specific spatial attention is both necessary and sufficient to induce learning. The spread of spatial attention, and thus of learning, is determined by the integrated effects of target distribution and task difficulty. We propose a theoretical framework whereby these factors combine to determine the cortical level of the focus of attention, which in turn enables learning modifications.

Macaque monkeys categorize images by their ordinal number.

The recall of a list of items in a serial order is a basic cognitive skill. However, it is unknown whether a list of arbitrary items is remembered by associations between sequential items or by associations between each item and its ordinal position. Here, to study the nonverbal strategies used for such memory tasks, we trained three macaque monkeys on a delayed sequence recall task. Thirty abstract images, divided into ten triplets, were presented repeatedly in fixed temporal order. On each trial the monkeys viewed three sequentially presented sample stimuli, followed by a test stimulus consisting of the same three images and a distractor image (chosen randomly from the remaining 27). The task was to touch the three images in their original order without touching the distractor. The most common error was touching the distractor when it had the same ordinal number (in its own triplet) as the correct image. Thus, the monkeys' natural tendency was to categorize images by their ordinal number. Additional, secondary strategies were used eventually to avoid the distractor images. These included memory of the sample images (working memory) and associations between sequence triplet members. Thus, monkeys use multiple mnemonic strategies according to their innate tendencies and the requirements of the task.

Learning pop-out detection: building representations for conflicting target-distractor relationships.

Studies of perceptual learning consistently found that improvement is stimulus specific. These findings were interpreted as indicating an early cortical learning site. In line with this interpretation, we consider two alternative hypotheses: the 'earliest modification' and the 'output-level modification' assumptions, which respectively assume that learning occurs within the earliest representation which is selective for the trained stimuli, or at cortical levels receiving its output. We studied performance in a pop-out task using light bar distractor elements of one orientation, and a target element rotated by 30 degrees (or 90 degrees). We tested the alternative hypotheses by examining pop-out learning through an initial training phase, a subsequent learning stage with swapped target and distracted orientations, and a final re-test with the originally trained stimuli. We found learning does not transfer across orientation swapping. However, following training with swapped orientations, a similar performance level is reached as with original orientations. That is, learning neither facilitates nor interferes to a substantial degree with subsequent performance with altered stimuli. Furthermore, this re-training does not hamper performance with the original trained stimuli. If training changed the earliest orientation selective representation (specializing it for performance of the particular performed task) it would necessarily affect performance with swapped orientations, as well. The co-existence of similar asymptotes for apparently conflicting stimulus sets refutes the 'earliest modification' hypothesis, supporting the alternative 'output level modification' hypothesis. We conclude that secondary cortical processing levels use outputs from the earliest orientation representation to compute higher order structures, promoting and improving successful task performance.

Task difficulty and the specificity of perceptual learning.

Practising simple visual tasks leads to a dramatic improvement in performing them. This learning is specific to the stimuli used for training. We show here that the degree of specificity depends on the difficulty of the training conditions. We find that the pattern of specificities maps onto the pattern of receptive field selectivities along the visual pathway. With easy conditions, learning generalizes across orientation and retinal position, matching the spatial generalization of higher visual areas. As task difficulty increases, learning becomes more specific with respect to both orientation and position, matching the fine spatial retinotopy exhibited by lower areas. Consequently, we enjoy the benefits of learning generalization when possible, and of fine grain but specific training when necessary. The dynamics of learning show a corresponding feature. Improvement begins with easy cases (when the subject is allowed long processing times) and only subsequently proceeds to harder cases. This learning cascade implies that easy conditions guide the learning of hard ones. Taken together, the specificity and dynamics suggest that learning proceeds as a countercurrent along the cortical hierarchy. Improvement begins at higher generalizing levels, which, in turn, direct harder-condition learning to the subdomain of their lower-level inputs. As predicted by this reverse hierarchy model, learning can be effective using only difficult trials, but on condition that learning onset has previously been enabled. A single prolonged presentation suffices to initiate learning. We call this single-encounter enabling effect 'eureka'.

Learning pop-out detection: specificities to stimulus characteristics.

Training induces dramatic improvement in the performance of pop-out detection. In this study, we examined the specificities of this improvement to stimulus characteristics. We found that learning is specific within basic visual dimensions: orientation, size and position. Accordingly, following training with one set of orientations, rotating target and distractors by 30 deg or more substantially hampers performance. Furthermore, rotation of either target or distractors alone greatly increases threshold. Learning is not transferred to reduced-size stimuli. Position specificity near fixation may be finer than 0.7 deg. On the other hand, learning transfers to the untrained eye, to expanded images, to mirror image transformations and to homologous positions across the midline (near fixation). Thus, learning must occur at a processing level which is early enough to maintain fine separability along basic stimulus dimensions, yet sufficiently high to manifest the described generalizations. We suggest that the site of early perceptual learning is one of the cortical areas which receive input from primary visual cortex, V1, and where top-down attentional control is present.

Perceptual stability and the selective adaptation of perceived and unperceived motion directions.

Adaptation was studied in a paradigm in which the adapting stimulus was a variably biased version of a bistable apparent motion stimulus, a motion quartet, and the post-adaptation test stimulus was a "neutral" motion quartet. Either horizontal or vertical motion was perceived, never both at the same time. When only one of these was perceived during the entire adaptation phase of a trial, and the perceived motion was highly stable, adaptation effects were greater for the perceived than the unperceived motion directions (i.e., adaptation was selective to the perceived motion). However, when the perceived motion during adaptation was relatively unstable (i.e., when the perceived motion was more likely to spontaneously change directions), similar levels of adaptation were obtained for perceived as well as unperceived, but possible motion directions. Thus, adaptation occurs prior to the determination of which of the competing motion directions will be perceived. The relationship between the stability of the adapting percept and the selectivity of adaptation is explained in terms of differences in the activation of mutually inhibitory horizontal and vertical motion detectors.

Restricted ability to recover three-dimensional global motion from one-dimensional motion signals: psychophysical observations.

We tested human ability to recover the 3D structure and motion information from time-varying images where only 1D motion cues were available. Under these conditions, observers exhibit poor performance in discriminating between two perpendicular axes of rotation, or discriminating between rigid and non-rigid 3D motion. This behavior of the visual system is to be contrasted with the good depth from motion performance exhibited when 2D motion cues are given in the image, as was found previously in numerous studies, and also in the work presented here. In a related paper, we suggest a theoretical framework in which to understand this differential performance on the basis of the two types of motion cues (1D vs 2D). Our findings are consistent with those of previous studies of frontoparallel motion, where it was shown that in many cases, the 1D cues alone were not integrated by the visual system into the correct global motion percept. This accumulating evidence suggests that oriented (1D) motion detectors alone cannot account for observed human performance of global motion perception, and that the role of units such as point or endpoint detectors should be studied further.

Restricted ability to recover three-dimensional global motion from one-dimensional local signals: theoretical observations.

Recovering 3D information from a 2D time-varying image is a vital task which human observers face daily. Numerous models exist which compute global 3D structure and motion on the basis of 2D local motion measurements of point-like elements. On the other hand, both experimental and computational research of early visual motion mechanisms emphasize the role of oriented (1D) detectors. Therefore, it is important to find out whether indeed 1D motion signals can serve as primary cues for 3D global motion computation. We have addressed this question by combining mathematical results and perceptual observations. We show that given the 2D-projected 1D instantaneous velocity field, it is mathematically impossible to discriminate rigid rotations from non-rigid transformations and/or to recover the rotation parameters. We relate this fact to existing results in cases where localized (point-like) cues are present, and to our own experiments on human performance in global motion perception when only 1D cues are given. Taken together, the data suggest a necessary role for localized information in early motion mechanisms and call for further physiological and psychophysical research in that direction.

Modeling perceptual learning with multiple interacting elements: a neural network model describing early visual perceptual learning.

We introduce a neural network model of an early visual cortical area, in order to understand better results of psychophysical experiments concerning perceptual learning during odd element (pop-out) detection tasks (Ahissar and Hochstein, 1993, 1994a). The model describes a network, composed of orientation selective units, arranged in a hypercolumn structure, with receptive field properties modeled from real monkey neurons. Odd element detection is a final pattern of activity with one (or a few) salient units active. The learning algorithm used was the Associative reward-penalty (Ar-p) algorithm of reinforcement learning (Barto and Anandan, 1985), following physiological data indicating the role of supervision in cortical plasticity. Simulations show that network performance improves dramatically as the weights of inter-unit connections reach a balance between lateral iso-orientation inhibition, and facilitation from neighboring neurons with different preferred orientations. The network is able to learn even from chance performance, and in the presence of a large amount of noise in the response function. As additional tests of the model, we conducted experiments with human subjects in order to examine learning strategy and test model predictions.

Isolating the effect of one-dimensional motion signals on the perceived direction of moving two-dimensional objects.

A considerable body of evidence suggests the existence of a two-stage mechanism for the detection of global motion. In the first stage the motion of elongated contours is extracted and then at the second stage these one-dimensional (1D) motion signals are combined. What is the nature of the computation carried out in combining the 1D motion signals towards forming a global motion percept? We devised a set of stimuli that differentiate between different possible computations. In particular, they distinguish between a velocity-space construction (such as intersection of constraints) and a linear computation such as vector averaging. In addition, these stimuli do not contain two-dimensional (2D) motion signals such as line intersections, that allow unambiguous determination of global velocity. Stimuli were presented in uncrossed disparity relative to the aperture through which they were presented, to reduce the effect of line terminator motion. We found that subjects are unable to detect the veridical global direction of motion for these stimuli. Instead, they perceive the stimulus pattern to be moving in a direction which reflects the average of its 1D motion components. Our results suggest that the visual system is not equipped with a mechanism implementing a velocity-space computation of global motion.

Attentional control of early perceptual learning.

The performance of adult humans in simple visual tasks improves dramatically with practice. This improvement is highly specific to basic attributes of the trained stimulus, suggesting that the underlying changes occur at low-level processing stages in the brain, where different orientations and spatial frequencies are handled by separate channels. We asked whether these practice effects are determined solely by activity in stimulus-driven mechanisms or whether high-level attentional mechanisms, which are linked to the perceptual task, might control the learning process. We found that practicing one task did not improve performance in an alternative task, even though both tasks used exactly the same visual stimuli but depended on different stimulus attributes (either orientation of local elements or global shape). Moreover, even when the experiment was designed so that the same responses were associated with the same stimuli (although subjects were instructed to attend to the attribute underlying one task), learning did not transfer from one task to the other. These results suggest that specific high-level attentional mechanisms, controlling changes at early visual processing levels, are essential in perceptual learning.

Asymmetric interactions in the processing of the visual dimensions of position, width, and contrast of bar stimuli.

The processing of different dimensions of a single stimulus may be either integral or separable. Dimensions are called integral if correlated variation of one improves discrimination on the basis of the other and random variation of one interferes with discrimination on the basis of the other. For separable dimensions on the other hand, subjects can attend to one dimension and disregard variations in the other. These discrimination tests were used to find the interactions between the processing of the visual dimensions of position, width, and contrast of a light bar stimulus. The relations between these dimensions were found to be asymmetric: judgments of position and width are independent of contrast variations, but variations in these dimensions influence contrast discriminations. Furthermore, position variations influence width judgements more than vice versa. The data were analyzed for repetition effects, and it was found that this model is not sufficient to explain all the interaction phenomena. The asymmetries found may be related to the different ways these dimensions are mapped onto cortical neuron responses.

Characteristics of the flux dimension cue in apparent motion correspondence.

It has been shown that element flux and size (but not luminance) serve as correspondence cues in the apparent motion visual system. Results are now presented of a study of the characteristics of the flux cue. It was found that flux rather than luminance is used by the system even when the size of the elements is greater than the size limit of Ricco's law. There were interactions between the apparent motion processing of the size and flux dimensions, beyond the obvious dependence of flux on size: positively correlated size and flux differences between elements have a greater effect on correspondence than do negatively correlated differences. Finally, when comparing the fluxes of different elements, the apparent motion system uses relative flux (above or below background) rather than absolute flux (relative to zero).

Gender differences in apparent motion perception.

Distance disparity is a strong cue to element correspondence in apparent motion. Using a 2-AFC paradigm we have previously shown that shape similarity also plays a role. We now demonstrate a small gender difference in these effects: women are more sensitive to distance disparity, whereas men are more sensitive to differences in shape. Furthermore, in the competing presence of a shape cue, women's sensitivity to distance decreases while men's sensitivity is unaffected. These observations may be related to putative gender differences in the 'form' and 'motion-spatial relations' cortical pathways.

On and off pathway contributions to apparent motion perception.

We studied the separability and/or interaction of the On and Off pathways in their role as inputs to visual motion perception. Using the long-range motion perception system, we asked if the motion system can use brightness polarity information, by testing whether correspondence is preferred between elements for which brightness polarity is preserved. We found such a preference, suggesting that brightness polarity information is indeed available to the motion system. However, under certain conditions motion is perceived even though the brightness polarity of apparent motion stimulus elements is reversed, indicating that the apparent motion system does integrate information from these two pathways. The source of the preference for maintaining polarity seems not to be the different brightnesses of the dark and bright stimulus elements, but the very fact that information must be integrated from different pathways. We relate the characteristics of the dependence of the motion perception on element contrast and contrast sign to those of previously reported visual evoked potential responses to brightness increments and decrements.

Time course of perceptual discrimination and single neuron reliability.

The reliability of identification of a visual target increases with time available for inspection of the stimulus. We suggest that the neural basis of this improvement is the existence of a mechanism for integrating a noisy firing rate over some period, leading to a reduction in mean firing rate variance with available processing time. We have determined the experimental time course of the improvement in reliability in a parallel search task where the available inspection time is limited by the presentation of a mask at various times after a brief stimulus. We compare the resulting psychometric functions with the predictions of a model based on Signal Detection Theory. The model is based on the assumption that the reliability of the observer's response is limited by the variability of the responses of individual neurons. The reliability of the discrimination between two stimuli at the neuronal level is then directly related to the ratio of the difference between their integrated mean responses (over many trials) to the response standard deviation. This reliability increases with inspection time. To demonstrate application of the model to electrophysiological data, "neurometric functions" are derived from the firing rates of a monkey V1 cortical neuron. The data were obtained while the animal was active in a discrimination task. The results correspond qualitatively to our observed human psychometric functions.

Visual orientation estimation.

A systematic error is reported in orientation estimation, in that on average, estimates are closer to the vertical axis than are the stimuli by up to 6 degrees. This systematic error results from a specific mechanism that may be related to depth perception, and that is avoided in certain circumstances or when other mechanisms take over. For example, the estimates of one observer who was a well-trained professional draughtsman did not show this systematic error. Furthermore, for all observers tested, estimation of clock time is not subject to the regular orientation estimation error. Rather, observers tend to estimate times as slightly further from the quarter hour than they really are. Orientation judgement channel capacity was also studied under various conditions. The number of discriminable orientations is far above the magic number "7" limit, reaching over 20 in optimal circumstances. The distribution of discriminable orientations is nonlinear, in that these are more closely packed about the horizontal and vertical axis than at the oblique.

Size, flux and luminance effects in the apparent motion correspondence process.

The effects of the relative size, luminance, and total luminous flux of apparent motion visual stimulus disk elements are studied, using a competitive paradigm. These dimensions can only be studied in pairs and we find that all three pairs have significant correspondence process effects. A comparison of the magnitudes of the effects, however, suggests that size and flux are the dimensions relevant to apparent motion processing, while luminance may not contribute to the correspondence process. Pitting distance against these dimensions in apparent motion tasks, we were able to find effective equivalence scales among them. Finally, interactions were found between the processing of some of these dimensions. The most pronounced interaction effect is that the addition of the size dimension increases the noise in the processing of distance, while size processing is not affected by the addition of the distance cue.