Range effects of an irrelevant dimension on classification.

In univariate classification tasks, subjects sort stimuli on the basis of the only attribute that varies. In orthogonal classification tasks, often called filtering tasks, there additionally are trial-to-trial variations in irrelevant attributes that the subjects are instructed to ignore. Performance is generally slower in filtering tasks than in univariate control tasks. We investigated this slowing in experiments of how the range of irrelevant trial-to-trial variation affects responses in pitch/loudness classification tasks. Using two levels of pitch and of loudness as stimuli, Experiment 1 replicated prior work showing that responses are slowed more when the range of the irrelevant dimension is made larger. Also in Experiment 1, sequential analyses showed that response time depends both on sequence and on the stimulus set independent of sequence. Experiments 2 and 3 used several levels on the irrelevant dimension and showed that responses to categorize loudness are slowed more by larger trial-to-trial pitch differences, but only on trials when the response repeats. When the response changes, performance is essentially unaffected by trial-to-trial irrelevant variation. This interaction supports the conclusion that slowed average performance in orthogonal classification tasks, which is known as Garner interference, is not due to difficulties that subjects have in filtering stimulus attributes. It is due to how subjects process successive stimulus differences. We call for more frequent reports of sequential analyses, because these can reveal information that is not available from data averages.

Psychophysical scaling: Judgments of attributes or objects?

Psychophysical scaling models of the form R = f(I), with R the response and I some intensity of an attribute, all assume that people judge the amounts of an attribute. With simple biases excepted, most also assume that judgments are independent of space, time, and features of the situation other than the one being judged. Many data support these ideas: Magnitude estimations of brightness (R) increase with luminance (I). Nevertheless, I argue that the general model is wrong. The stabilized retinal image literature shows that nothing is seen if light does not change over time. The classification literature shows that dimensions often combine to produce emergent properties that cannot be described by the elements in the stimulus. These and other effects cannot be adjusted for by simply adding variables to the general model because some factors do not combine linearly. The proposed alternative is that people initially judge the entire stimulus - the object in terms of its environment. This agrees with the constancy literature that shows that objects and their attributes are identified through their relations to other aspects of the scene. That the environment determines judgments is masked in scaling studies where the standard procedure is to hold context constant. In a typical brightness study (where different lights are presented on the same background on different trials) the essential stimulus might be the intensity of the light or a difference between the light and the background. The two are perfectly confounded. This issue is examined in the case of audition. Judgments of the loudness of a tone depend on how much that tone differs from the previous tone in both pitch and loudness. To judge loudness (and other attributes) people first seem to process the stimulus object in terms of differences between it and other aspects in the situation; only then do they assess the feature of interest. Psychophysical judgments will therefore be better interpreted by theories of attention that are based in biology or psychology than those (following Fechner) that are based in classical physics.

Toying with the moon illusion.

We propose that the correct interpretation of the moon illusion is that the zenith moon appears small, not that the horizon moon appears large. This illusion is caused by the visual gap between the observer and the overhead moon. Because of the gap, the observer has no or little optical information about the distance of the moon. This results in empty field myopia where the moon is neurally, although not necessarily cognitively, processed as being at about arm's length. When the moon is seen on the horizon, there usually is optical information about distance. That results in reduced accommodation, and so the moon is processed as at a greater distance. Consistent with the size-distance-invariance hypothesis, the moon is then judged as large. This is a specific example of the more general fact that all distant objects appear small in the absence of a stimulus for accommodation to be distant. This outcome produces the toy illusion.

McCollough aftereffects in strabismus and amblyopia.

In strabismic people, the functioning of the deviating eye is impaired. This impairment is binocular in origin since it depends on stimulation of the straight eye; sensitivity and acuity of the deviating eye are usually better when the straight eye is patched than when both eyes are viewing. We investigated interactions between the eyes in strabismic people by using the McCollough color aftereffect as an exploratory procedure. Both normal and strabismic people adapted to black and colored bars and then examined black and white test bars. They reported colors in the achromatic test bars that were complementary to the adapting colors; this is the McCollough aftereffect. Both a monocular McCollough effect and a binocular McCollough effect were induced in people having normal vision. Strabismic people did not show a binocular McCollough effect. Aftereffect strengths were the same in each eye of normal people but were stronger in the impaired eye of strabismic people. A speculation is offered as to why the "non-seeing" eye of strabismic people "sees" the aftereffect better.

Incidence and characteristics of McCollough aftereffects following video display terminal use.

After using video display terminals (VDT), some persons notice that achromatic patterns appear faintly colored hours after terminal use. We investigated the incidence of this effect, the McCollough effect (ME), among 125 VDT users. Subjects completed a questionnaire regarding work habits and certain life-style aspects. They were shown photographs of varying spatial frequencies to identify the adapting stimulus responsible for the ME. The incidence of ME after routine use of VDTs was 19.1%. The adapting stimulus was the repeating character lines of the VDT. No aspect of life-style investigated appeared to predispose an individual to develop the ME. Those subjects who developed the ME did not differ from those who did not in age, sleep, caffeine consumption, use of medication, refractive errors, or computer usage, nor did they have a higher incidence of ocular defects or eye strain.

Brightness of luminance distributions with gradual changes.

The perception of gradually changing luminance distributions was investigated. Luminance changed across the radius of a disk by a linear, quadratic, or cubic function with varying magnitudes. Subjects selected matching luminances for the inner and outer edges of each stimulus. The threshold for reporting that the inner and outer matches were different occurred at approximately 20% contrast between those regions. This threshold did not vary with the particular function which described the luminance distribution. Further, as the magnitude of luminance change across the stimulus increased, subjects judged the inner and outer edges to differ more in brightness. Matching luminances also depended upon the background configuration with greater differences perceived across the disk radius when the surround and center dot of the stimulus were of opposite, rather than the same, luminance. These results indicate that models of brightness must consider all luminance changes in the stimulus, not just changes of a particular type such as second differences or changes at a luminance step.

Momentary and global maximizing.

Four adult females responded at a computer console, on three constant probability concurrent variable-interval reinforcement schedules. The subjects were instructed to try to obtain as many reinforcers as possible, but were not given any instructions on how to accomplish this task. Three of the four subjects typically allocated responses to the schedule offering the higher reinforcement probability. These results show that some subjects trying to maximize overall reinforcement may respond in accordance with a momentary maximizing strategy.

Range effects in successive discrimination.

In two sets of experiments we examined pigeons' discrimination performance with a visual flicker-rate continuum, using a conventional successive discrimination procedure. In the first experiment, responses during the intermediate stimulus value were never reinforced, while responses during stimuli on either end of the continuum were reinforced periodically. In the second experiment, responses during stimuli from one end of the continuum were never reinforced, while responses during stimuli from the other end of the continuum were reinforced periodically. Results from both experiments show that discrimination between unchanged positive and negative stimulus values is a function of the range over which the total stimulus set varies. These range effects are comparable to effects found in absolute judgment tasks in human and animal psychophysics. In addition, the range effects reported here are not due to channel capacity, but may depend instead on variability in judgment criteria.

A memory model of sequential effects in scaling tasks.

Subjects judge successive stimuli to be overly similar in psychophysical scaling tasks. This is called assimilation. They also tend to judge each stimulus as overly different from more previous events. This is called contrast. To examine a two-stage linear model of these sequence effects, we asked subjects to judge the relative intensity of successive tones. In support of the model, responses again depended lawfully on prior events. These memory effects occur in a variety of scaling tasks and are consistent with two assumptions: (a)Successive events assimilate in memory, and (b) subjects compare each stimulus to a collection of memories of prior events to generate a response. The trial-by-trial analysis used to test the model also showed that even in magnitude-estimation studies, equal stimulus ratios do not result in equal response ratios, except on average. This article suggests that examinations of trial-by-trial performance might be useful in studying memory and judgment processes.

Holistic versus analytic process models: a reply.

Two classes of stimulus process models are considered in this reply to Dykes and Cooper. It is shown that analytic models which assume that stimuli are initially processed in terms of constituent dimensions do not account for large amounts of published data. It is also shown that the holistic-discriminability model that Dykes and Cooper reject is nonetheless consistent with their results and predicts all of the data for which their analytic model was constructed to account.

Classifying intergral stimuli.

Two reported experiments support holistic, as opposed to analytic, processing models for integral stimuli. Speeded classification data from different information-processing tasks (univariate and correlated) were predicted by distance between stimuli in similarity space but not by redundancy. The results of the filtering and condensation tasks and the notion of configural stimuli are also explicable in these terms. It is shown that some operational definitions commonly used to define integral stimuli are usually confounded with stimulus similarity. The assumption of independence between the attributes that combine to form multidimensional stimuli is not always met and is always an empirical question. When these attributes are not independent, physical and psychological spaces are not necessarily the same. Similarity structure is a crucial concern if inferences of cognitive processing are to be based on information-processing task results.

Parameters of stimulation and perception in an artificial sensory feedback system.

The relationships between stimulus parameters and perceptions in a prosthetic feedback system were measured using psychophysical methods. Electrical stimulation of the median nerve produced a monotonic relation between frequency of stimulation and the perceived magnitude of the stimulus. There were two qualitatively different perceptions of the stimulation; one for low frequencies and one for high. These two qualities fit different psychophysical continuua, kind of stimulation, and amount of stimulation.