A Conversation with Jacob Nachmias.

We are sad to report that Professor Jacob (Jack) Nachmias passed away on March 2, 2019. Nachmias was born in Athens, Greece, on June 9, 1928. To escape the Nazis, he and his family came to the United States in 1939. He received his undergraduate degree from Cornell University and then an MA from Swarthmore College, where he worked with Hans Wallach and Wolfgang Kohler; his PhD in Psychology was from Harvard University. Nachmias spent the majority of his career as a Professor of Psychology at the University of Pennsylvania. He made fundamental contributions to our understanding of vision, most notably through the study of eye movements, the development of signal detection theory and forced-choice psychophysical methods, and the psychophysical characterization of spatial-frequency-selective visual channels. Nachmias' work was recognized by his election to the National Academy of Sciences and receipt of the Optical Society's Tillyer Award.

Shape and size discrimination compared.

Observers presented with pairs of figures differing in area (SIZE) or aspect ratio (SHAPE) spontaneously make use of both height and width differences. whether or not they are forced to do so by between-interval jittering or even instructed to do so. SHAPE discrimination is considerably better than SIZE discrimination. The superiority of SHAPE discrimination is probably due to partial correlation between the encoding noise of height and width of a figure. Discrimination of height differences is seemingly increased (decreased) by negatively (positively) correlated width differences, relative to leaving width unchanged. This is true whether the different types of trials are presented in separate blocks or intermixed. Perhaps SIZE and SHAPE comparisons are always made and their decision variables are optimally combined. The difference between SIZE and SHAPE discrimination is reduced, if not reversed, when figures are presented simultaneously rather than successively. This interaction between type of task and mode of presentation, may be due to the increased amount of correlation between test and standard figures of the encoding noise common to the two dimensions of each figure.

Judging spatial properties of simple figures.

Discrimination was measured for height, area, and aspect ratio of ovals and rectangles. Random jittering of the orthogonal property (width, aspect ratio, and area) was used to control the observers' criterion. Weber fractions for aspect ratio were consistently lower than those for area, and about the same as those for height. Performance with ovals and rectangles did not differ significantly. Two different methods were employed to assess the side effects of jittering. It was found that jittering reduces the discriminability of each property, though less for aspect ratio than for height or area. The hypothesis that judgements of both area and aspect ratio are linear combination of noisy estimates of height and width predicts Weber fractions for aspect ratio and for area to be 2(1/2) times higher than those for height. Results from unjittered trials clearly reject the hypothesis with respect to aspect ratio but not for area.

The role of virtual standards in visual discrimination.

Psychometric functions were measured in various visual discrimination tasks involving test stimuli whose values spanned a standard stimulus. In two-interval trial blocks, the standard was either always in the first or always in the second interval, or appeared randomly in either interval. In one-interval blocks, the standard stimulus was never presented. Fitting the data with cumulative Gaussian functions revealed that discriminability was highest on one interval trials, where the observer had to rely on an implicit standard. On two-interval trials, discriminability was higher when the standard was in the first rather than the second interval, regardless of whether those two types of trials were intermixed or not, also possibly implicating the operation of an implicit standard in two-interval trials as well. A time-order error occurs on two-interval trials: in effect the value of the stimulus presented in the first interval is underestimated relative to that in the second interval. An analogous error occurs in one-interval trials, as if there were an implicit standard whose value is underestimated.

Contrast discrimination with and without spatial uncertainty.

The main purpose of these experiments was to examine in detail how successfully the uncertainty reduction explanation accounts for the effects of spatial cues in a temporal forced-choice contrast discrimination task in which any one of four well-separated Gabor patches is incremented. In preliminary experiments, it was shown that in the absence of uncorrelated contrast jitter, observers could use either of at least two different decision strategies, whereas in the presence of random contrast jitter, their response was based on the largest contrast value in the two intervals of a trial. Independent spatial cues were used in each interval to minimize the likelihood of eye movement artifacts. Cues either preceded each stimulus presentation by 100 ms, or were coincident with it. Precues were only slightly more effective than simultaneous cues, and either of them improved performance nearly as much as could be expected from the uncertainty reduction account of spatial attention.