Some additional predictions and further tests of the Marr-Ullman model of motion perception

The Marr-Ullman model for motion detection in the human visual system functions by means of the dual input of polarity-specific edge detectors and luminance change detectors. Moulden and Begg (1986) found a polarity-specific motion aftereffect which they claimed provided support for this dual input model. The logic of their experiment is examined, and it is shown that several additional predictions arise from the Marr-Ullman model, which were not supported by Moulden and Begg's study. A more powerful experiment was carried out and these additional predictions were disconfirmed, although the polarity-specific effect did emerge. A consideration of alternative explanations of this effect led to a second experiment in which an attempt was made to discover the actual determinants of the effect. This revealed that polarity-specific units are unlikely to play any part in the phenomenon. It was concluded, in the light of this and other evidence, that one of a class of alternative models is more likely to be the actual mechanism for motion perception. However, careful consideration of the Marr-Ullman model indicated that it may be untestable in principle if various differentially weighted levels of neural integration are envisaged.     

Some tests of the interactive-activation model for word identification

Summary The interactive-activation model postulates (a) that activation at the letter level leads automatically to activation at the word level, (b) that the word-superiority effect reflects reactivation of letters by the word they spell, and (c) that subjects identify words on the basis of information obtained from separate letter-position channels. In the first two experiments, we showed words in upper, lower, or mixed case: the word-superiority effect was reduced when words were presented in mixed-case letters, presumably because extra-letter information is lost with mixed-case presentation; i.e., postulate (c) is wrong. The third experiment showed that when the letters of a word are rotated 180° subjects can identify the letters without producing a word-superiority effect; i.e., one of postulates (a) and (b) is wrong. In Experiments 4 and 5, we trained subjects to name words presented in inverted letters; training was more effective when subjects could exploit bigram information in addition to letter-channel information; i.e., reading inverted text is based on extra-letter-feature information, not on a general skill in rotating letters. Taken together, our data deny three of the interactive-activation model's major postulates. We offer some suggestions for future versions of the model. Electronic mail: Userid: MEWHORTD; Nodeid: QUCDN; Domain: BITNET     

In defence of a ratio model for movement detection at threshold

Sutherland (1961) proposed that the detection of motion might depend upon the ratios of firings in cells sensitive to movement in opposite directions. Sekuler and his collaborators have argued that the notion of a ratio mechanism at threshold is wrong. The findings and arguments upon which this conclusion was based are challenged, an explicit model is described which provides an account of data previously held to be inconsistent with a ratio model, and an experiment is reported which provides unequivocal support for the ratio model and whose findings are inconsistent with the predictions from Sekuler's “independence” model.     

Apparent movement detection in the pigeon

Pigeons were trained to discriminate between the presence and absence of apparent movement in visual displays. Generalization gradients obtained on the dimension of speed indicated a broad range of movement detection. Extremely accurate discrimination performance on displays varying in direction of movement suggested formation of a "movement concept".     

Some further tests of the constant-ratio rule

Five experiments sought to test the constant-ratio rule (CRR) with single dimensional ensembles composed of 2, 4, or 8 stimulus objects. Each S attempted to identify stimuli which varied in weight or in visual size or brightness. The results demonstrated: (a) The CRR predicts equally well the response proportions of single dimensional visual, kinesthetic, and auditory stimulus ensembles, but less well than those for multidimensional auditory stimuli. (b) Better predictions are obtained with four than with two stimulus objects.(c) The CRR is sensitive to variations in the spacing and range of the stimulus ensembles and to practice on the task. It is concluded that the rule tends to fail whenever stimulus conditions elicit differential amounts of stimulus and response confusion.     

Some tests of the decay theory of immediate memory

The hypothesis of decay of the memory trace as a cause of forgetting has been unpopular. The reasons for this unpopularity are criticized and a theory of the memory span, based on this hypothesis, is put forward. Three experiments which test the hypothesis are described. In each, two kinds of stimuli are presented to the subject, viz., “required” stimuli, which he attempts to remember, and “additional” stimuli, to which he merely makes responses. The first experiment will show that even when the number of required stimuli is well below the memory span, forgetting occurs if the presentation of additional stimuli delays recall for several seconds. The second shows that the effect of the additional stimuli depends only slightly on their similarity to the required stimuli: it also shows that their effect is negligible when they precede, instead of follow, the required stimuli. The third shows that the effect of additional stimuli interpolated before recall remains considerable even when there is an interval of several seconds between presentation of required and additional stimuli.     

Loglinear Rasch model tests

Existing statistical tests for the fit of the Rasch model have been criticized, because they are only sensitive to specific violations of its assumptions. Contingency table methods using loglinear models have been used to test various psychometric models. In this paper, the assumptions of the Rasch model are discussed and the Rasch model is reformulated as a quasi-independence model. The model is a quasi-loglinear model for the incomplete subgroup × score × item 1 × item 2 × ... × itemk contingency table. Using ordinary contingency table methods the Rasch model can be tested generally or against less restrictive quasi-loglinear models to investigate specific violations of its assumptions.     

Nonparametric goodness-of-fit tests for the rasch model

A Monte Carlo algorithm realizing a family of nonparametric tests for the Rasch model is introduced which are conditional on the item and subject marginals. The algorithm is based on random changes of elements of data matrices without changing the marginals; most powerful tests against all alternative hypotheses are given for which a monotone characteristic may be computed from the data matrix; alternatives may also be composed. Computation times are long, but exactp-values are approximated with the quality of approximation only depending on calculation time, but not on the number of persons. The power and the flexibility of the procedure is demonstrated by means of an empirical example where, among others, indicators for increased item similarities, the existence of subscales, violations of sufficiency of the raw score as well as learning processes were found. Many of the features described are implemented in the program T-Rasch 1.0 by Ponocny and Ponocny-Seliger (1999).The author wishes to thank Alexander Kaba, Birgit Bukasa, and Ulrike Wenninger of Österreichisches Kuratorium für Verkehrssicherheit (Austrian Traffic Safety Board) for allowing a data set to be used for the empirical example, and Elisabeth Ponocny-Seliger and the reviewers for many helpful comments. The menu-driven program T-Rasch 1.0 by Ponocny and Ponocny-Seliger (1999) can be obtained from Assessment Systems Corporation (http: //www.assess.com) or from the authors. (Note that it also performs exact person fit tests.)     

Some approximate tests for repeated measurement designs

Four approximate tests are considered for repeated measurement designs in which observations are multivariate normal with arbitrary covariance matrices. In these tests traditional within-subject mean square ratios are compared with critical values derived fromF distributions with adjusted degrees of freedom. Two of them—the approximate and the improved general approximate (IGA) tests—behave adequately in terms of Type I error. Generally, the IGA test functions better than the approximate test, however the latter involves less computations. In regards to power, the IGA test may compete with one multivariate procedure when the assumptions of the latter are tenable.The author wishes to thank Garrett K. Mandeville for his careful reading of the final version of the paper.     

Some factors affecting response in recognition tests

Experiments were designed to investigate some of the factors which might affect a subject's responses in a recognition test. Two experiments are described. In the first, subjects were shown 12 similar photographs. They were then given a recognition test of 12 photographs, of which a number varying from o to 12 had been shown in the original display. All the photographs were sufficiently alike to render some guessing essential. It was found that subjects tended to select approximately 50 per cent. as having been seen previously and to reject 50 per cent. When a second and similar display and recognition test followed immediately, there was a tendency for subjects to make more “yes” responses, without any improvement in accuracy. The effect of giving serial presentation first was to increase the number accepted not only in that test but also in a test using simultaneous presentation which succeeded it. It was also shown that when a correct decision was possible but difficult a positively correct response (in this case recognizing something seen before) was easier than a negatively correct one (rejecting something not seen before).

In the second experiment subjects were shown a picture and were then asked to describe it from memory. Immediately afterwards half the group was given a multiple-choice recognition test while the other half was given a single-choice recognition test. It was found that accuracy and certainty of decision varied with the nature of the choice situation. The results are discussed in relation to a theory of adaptation levels.     

The semantic congruity effect: Some tests of the expectancy hypothesis

Subjects comparing items in memory along some dimension are usually quicker to specify the lesser (than the greater) of two low magnitude items and the greater (than the lesser) of two high magnitude ones. One account explains this congruity effect as due to subjects instructed to specify the higher as expecting high magnitude items to follow and the reverse being true for subjects specifying the lesser. Three experiments tested this expectancy hypothesis. In experiment 1, subjects were set to the actual size range of each pair before the pair was shown but the congruity effect still occurred. In experiments 2 and 3, subjects compared critical pairs from a narrow size range plus more from either the same or much broader ranges. Times to compare the critical pairs were the same regardless of the range of the other pairs that subjects were exposed to. These results are strong evidence against the expectancy hypothesis.     

Experimental tests of the mean-variance model for portfolio selection

Statistically knowledgeable male and female undergraduate students participated in 40 portfolio selection problems with monetary payoff contingent on performance. The portfolio selection task included two independent risky assets with normally distributed returns. It provided access to information about previous returns, allowed borrowing and lending at a fixed interest rate, and forced on each decision period a choice between the two risky assets. The findings show a high percentage of inefficient mean-variance portfolios which does not decrease with practice, a high rate of requests for useless information, a large frequency of switches between the two risky assets, and sequential dependencies. Theoretical and practical implications of the findings are briefly discussed.