Feature analysis and the growth of a percept.

Twelve trigrams divided into four sets, the members of each set varying systematically in the feature configuration at each letter position, were repeatedly exposed at brief durations in a random order to subjects who were asked to report after each exposure those letters that could be recognized with certainty. The set of three trigrams that had the same features for all letters in a given ordinal position was identified with fewer exposures than were necessary to permit identification of sets for which the featural characteristics varied within a spatial position. The results were interpreted as being consistent with a constructive model of letter recognition in which the observer tests hypotheses regarding the presence of a given letter based on the presence of relevant features in a spatial position.

Positional differences in performance on members of confusable and nonconfusable letter pairs.

A parafoveally presented letter is more accurately identified when flanked by a letter to its foveal side than when flanked by one to its peripheral side, but only if the two letters are nonconfusable. With confusable letters there is no such relative position effect. Four experiments indicated that the basis for this confusability-asymmetry interaction is not criterion or response bias, but rather that it occurs earlier in visual processing. In Experiment 1 the interaction was found when only one pair member was reported, thus eliminating response bias requiring the report of both letters as the source of the effect. In Experiment 2 the data were subjected to signal detection analysis, and the interaction persisted. In Experiment 3 pair members were presented simultaneously or in rapid sequence, and the interaction was found only with simultaneous presentations. In Experiment 4 letters were used with upper- and lowercase counterparts that were quite different in shape. Uppercase letters that were most and least confusable for each subject were paired for presentation in their upper case form or in mixed-case form. The interaction occurred only with upper-case pairs.

Facilitatory or Inhibitory Nontarget Effects in the Location-Cuing Paradigm

The effect of nontargets on the identification of targets in the location-cuing paradigm was investigated in order to determine whether observers consistently allocate their attention to a validly cued location and whether the effect of nontargets is to facilitate or to inhibit performance. In four experiments, the effects of a single matching nontarget or a single nonmatching nontarget were compared. In each experiment, it was shown that observers consistently allocate their attention to a cued location when a precue appears and that performance is inhibited more by nonmatching nontargets in the display than it is facilitated by matching nontargets.

Is rapid performance improvement across short precue-target delays due to masking from peripheral precues?

The role of lateral masking in more rapid performance improvement with peripheral than with central precuing was investigated. A peripheral precue to the inside of the target location provided less masking at zero precue-target delay than a precue to the outside (experiment 1) or a precue involving a partial target at the target location (experiment 2). There was no significant interaction between precue-target delay and precue type in a comparison of inside precues and precues involving a briefly-brightened box around the target location, although overall performance was significantly poorer with the latter (experiment 3). Performance was better at short precue-target delays with inside precues than with central precues (experiment 4), yet it did not improve significantly more rapidly. Minimizing lateral masking with peripheral precues thus eliminates the dramatic performance improvement sometimes observed across short precue-target delays, causing performance to be consistently better than with central precues across these delays.

Effects of relative position and shape relation on gap detection with geometric form pairs.

A letter presented parafoveally is better identified when flanked by a letter to its foveal side than when flanked by a letter to its peripheral side if the letters are different in shape. If the letters are similar, relative position does not affect identifiability. Whereas some articles in the literature suggest that this effect may not extend to geometric forms, results of the first experiment showed that it does. Results of the second experiment revealed that the strength of the significant interaction between relative position and shape relation remains constant over a range of eccentricities (2-20 deg were tested) extending above and below as well as to the left and right of fixation. Results are compared with predictions from a model that can account for the relative position-shape relation interaction.

Time course of attention effects with abrupt-onset and offset single- and multiple-element precues.

Large differences between the time course of attentional responsiveness to onset single-element precues (onset singles) and to onset multiple-element precues (onset multiples) have suggested differences in the way attention is controlled. In five experiments here, singles presented as offsets produced rapid attention buildup, attentional decay across longer precue-to-target delays, and attentional capture, as do onset singles, suggesting exogenous attentional control; both offset and onset multiples produced gradual onset of attentional effects without subsequent attentional decay, suggesting endogenous attentional control; and onset and offset singles produced higher accuracy than onset and offset multiples. Thus, the dynamic quality of a sudden onset is not sufficient explanation for the exogenous attentional control produced by a single-element peripheral precue.

Effects of abruptly appearing clutter on a peripherally precued covert attention shift.

The effects of clutter that appears abruptly with the target during covert shifts of attention to a precued target location in the visual field were studied. The delay between the appearance of the precue and that of the target with clutter was varied within each block. Blocked conditions included clutter density, the similarity between the shape of the clutter character and that of the target, and whether a clutter character appeared in each path between fixation and the target. Also, a no-clutter control condition was run in this totally within-subject design. Implications of the results for early versus late-selection models of attention were assessed.

Visually-presented letter strings typically are encoded phonologically: some converging evidence.

Three experiments were conducted to show that phonological encoding is typical for visually-presented letter strings, and that an interactive activation model with a phonological route to the mental lexicon accounts adequately for the word-superiority effect. In Experiment 1, pseudohomophones produced a word-superiority effect that was as great as that produced by words. More accurate target discrimination in homophones than in nonhomophones in Experiment 2 was interpreted to mean that excitability of entries in the mental lexicon increases with frequency of access. Target discrimination accuracy was inversely related to the phonological complexity of strings containing targets in Experiment 3, supposedly because lexical access through which target discrimination is enhanced becomes more difficult as phonological complexity increases.

Does order of letter analysis contribute to the parafoveal identification asymmetry?

A parafoveally presented target character usually is identified more accurately when flanked by a nontarget character to its foveal side than when flanked by one to its peripheral side. An outside-in process of analysis produced by uncertainty about the target's position could contribute to this asymmetry. Current results revealed a greater asymmetry with relative target position blocked than with it mixed over trials, suggesting that target position uncertainty leads to inside-out, rather than outside-in, analysis.

Target letter inclusion in single and multiple word substrings and identification accuracy.

In experiment 1, target letter discrimination in quadrigrams containing multiple subwords was found to be no more accurate than target letter discrimination in strings of four unrelated letters. Discrimination in quadrigrams, each including the target letter and a one-word trigram of which the target was not a part, was significantly poorer than discrimination in either of the other two types. In Experiment 2, a possible explanation of the latter result involving capacity limitations was eliminated by presenting the target above or below the other letters, which were presented in the same order as in corresponding strings in Experiment 1. The results are explained with respect to the interactive activation model of context effects in letter perception, with a modification proposed involving inhibitory connections between word nodes.

Facilitation in letter detection with sequentially presented homophonic strings.

A consistent phonology between sequentially presented pseudoword quadrigrams was found to improve performance in a letter-detection task. Each of 8 subjects was required to judge whether an M appeared in either the initial or terminal position of letter strings. The letter N was used as a foil. For each presentation, two quadrigrams appeared sequentially, the first for 20 ms and the second for a duration that allowed approximately 75% overall accuracy. The strings differed only by the letter next to the one on the opposite end of the string from the M or N. Subjects were not told that two strings were exposed on each presentation, and when asked, none reported having been aware of this. Nevertheless, detection was significantly better when the two strings were homophones (e.g., SERM, SIRM) than when they were not (e.g., SARM, SORM). Results are discussed in terms of phonological access to the mental lexicon.

Time-course of sensitivity changes as attention shifts to an unpredictable location.

Shifting visual attention is often described as analogous to a spotlight moving through empty space between locations. In the present experiment, a peripheral precue summoned attention to an initial location, and 200 ms later a second peripheral cue appeared beside one of two possible second locations, each 14 degrees away from the initial location. The target was twice as likely to appear at the location that had been indicated by the immediately preceding cue as at the location that had been previously cued or that would be cued. Fine-grained temporal analyses indicated that, as attention was shifted, sensitivity to information at the second location gradually increased while sensitivity at the first location simultaneously decreased. Average sensitivity over the two locations during the shift remained significantly greater than average sensitivity immediately following the initial precue. In contrast, an attentional "spotlight" moving from the first to the second location would produce a decrease in average sensitivity to its initial level while the spotlight was between locations.

Parafoveal identification asymmetry in a lateral masking effect.

The degree to which the identification of a parafoveally presented letter target was impaired by being flanked by a foveal or peripheral nontarget at a target-mask interval of zero, 75, or 150 msec, was examined. Peripheral placement of the nontarget was more disruptive than foveal placement, and this asymmetry was significantly more pronounced at a 75-msec interval than at either of the other two. This finding is consonant with explanations of the asymmetry based on target visibility rather than on criterion or response factors.

Phonological access to the mental lexicon in a target discrimination task.

A phonological route to the mental lexicon was demonstrated in two experiments. Each involved a target discrimination task in which the target was the first or last letter of a quadrigram. Half the quadrigrams were pseudohomophones (identical in sound but not in spelling to an English word) and the other half were nonhomophones (pronounceable but neither spelled like nor sounding like any English word). In Experiment 1 each pseudohomophone had a nonhomophone counterpart produced by changing only the letter on the opposite end of the string from the target. In Experiment 2 the nonhomophone counterparts were produced by changing only the letter next to the one on the opposite end of the string from the target; thus the extreme letters were the same for both types of quadrigrams. Performance was significantly better on pseudohomophones than on nonhomophones in both experiments.

A test of perceptual enhancement with peripherally presented letter pairs.

The present study was conducted to determine whether sensitivity to a member of a pair of stimulus letters projecting to the parafovae would be improved by presenting that member at fixation. The effect was examined for the pair member appearing at a constant distance from fixation, which always was flanked to its inner or outer side. Contrary to expectations based upon results reported by Geiger and Lettvin (1986), observed accuracy changes were found to be due to response bias rather than to sensitivity enhancement.

Representation of letters when mislocation errors occur.

Errors in reporting a cued target letter appearing among a string of letters more often reflect the mislocation of a letter appearing elsewhere in the string than the intrusion of one not in the string. The current experiment was conducted to determine the representation of letters at the stage at which errors occur. Four letters (from a set of 12 chosen to contain counterpart pairs that were similar physically, phonemically, or both physically and phonemically) appeared in each exposure, with the target letter indicated by a cue in the postexposure mask. Letter strings presented to one group of subjects were flanked on either side by a pound sign (#) to assess the effect of lateral masking on the terminal letter in the string. Physical similarity dictated the pattern of mislocations between counterparts, suggesting a physical rather than phonemic or abstract representation. Lateral masking played no significant role in the difference between intrusions and mislocations.

When shape information from a foveal nontarget affects gap detection accuracy in a parafoveal target.

Effects on gap detection accuracy of information at fixation concerning a parafoveally presented target form's global shape were determined in three experiments. Shape information was not directly relevant to the correct response regarding the presence of the gap. Tests were conducted to determine whether costs and benefits were associated with shape information, whether the effectiveness of shape information was sensitive to spatial attention, and whether subjects could volitionally suppress the nontarget information if it became apparent from preattentive analysis that it would not be helpful. Results were positive in all instances. Conclusions were that the foveal information was subject to early selection and that therefore it was not processed automatically.

Inhibition of return: support for generality of the phenomenon.

Inhibition of return (IOR), first described in 1984, was considered to be a general phenomenon for ensuring that attention would be allocated to successive stimuli in the environment. In the present research, IOR was expressed in forced-choice identification tasks with either reaction time or accuracy as the dependent measure. Thus, the generality of IOR was supported, because response inhibition cannot explain IOR found with accuracy measures. Concepts from the variable and permeable filters metaphor are used to suggest how changes in attention can change expression of IOR by rapid variation in perceptual threshold.

Asymmetric parafoveal interference with dichoptic presentations.

A Landolt C target and letter X nontarget were dichoptically presented to 8 subjects. The nontarget presented to one eye foveally or peripherally flanked the area of the parafovea corresponding to that receiving the target in the other eye. Identification of the orientation of the C was poorer with the nontarget in the peripheral position. If, as has been proposed previously, asymmetric lateral inhibition is responsible for the phenomenon, its locus would seem to lie beyond the retina.

Parafoveal identification asymmetry: interactive effects of shape and color.

If two adjacent letters project to the parafoveal region of the retina, both accuracy and discriminability measures have revealed that a letter flanked to its foveal side is identified more accurately than a letter the same distance from the fovea that is flanked to its peripheral side. This parafoveal identification asymmetry is greater if the letters are dissimilar in shape than if they are similar. Color and brightness were introduced as variables in the present experiments. The identification asymmetry was greatest for dissimilar letters in different (complementary) colors. Although those colors differed also in brightness, two letters that were achromatic but merely different in brightness did not produce an asymmetry interaction with shape. Interletter separation was varied between .15 and 1.95 deg, and the pattern of results just described persisted across both distances. The synergistic interaction of shape relation and color relation in determining the amount of identification asymmetry suggests that color and shape affect perceptual processing at the same level.