The contents of visual working memory reduce uncertainty during visual search

Information held in visual working memory (VWM) influences the allocation of attention during visual search, with targets matching the contents of VWM receiving processing benefits over those that do not. Such an effect could arise from multiple mechanisms: First, it is possible that the contents of working memory enhance the perceptual representation of the target. Alternatively, it is possible that when a target is presented among distractor items, the contents of working memory operate postperceptually to reduce uncertainty about the location of the target. In both cases, a match between the contents of VWM and the target should lead to facilitated processing. However, each effect makes distinct predictions regarding set-size manipulations; whereas perceptual enhancement accounts predict processing benefits regardless of set size, uncertainty reduction accounts predict benefits only with set sizes larger than 1, when there is uncertainty regarding the target location. In the present study, in which briefly presented, masked targets were presented in isolation, there was a negligible effect of the information held in VWM on target discrimination. However, in displays containing multiple masked items, information held in VWM strongly affected target discrimination. These results argue that working memory representations act at a postperceptual level to reduce uncertainty during visual search.     

Iconic memory for natural scenes: Evidence using a modified change-detection procedure

ABSTRACT

Change blindness for the contents of natural scenes suggests that only items that are attended while the scene is still visible are stored, leading some to characterize our visual experiences as sparse. Experiments on iconic memory for arrays of discrete symbols or objects, however, indicate observers have access to more visual information for at least several hundred milliseconds at offset of a display. In the experiment presented here, we demonstrate an iconic memory for complex natural or real-world scenes. Using a modified change detection task in which to-be changed objects are cued at offset of the scene, we show that more information from a natural scene is briefly stored than change blindness predicts and more than is contained in visual short-term memory. In our experiment, a cue appearing 0, 300, or 1000?msec after offset of the pre-change scene or at onset of the second scene presentation (a Post Cue) directed attention to the location of a possible change. Compared to a no-cue condition, subjects were significantly better at detecting changes and identifying what changed in the cue condition, with the cue having a diminishing effect as a function of time and no effect when its onset coincided with that of the second scene presentation. The results suggest that an iconic memory of a natural scene exists for at least 1000?msec after scene offset, from which subjects can access the identity of items in the pre-change scene. This implies that change blindness underestimates the amount of information available to the visual system from a brief glance of a natural scene.     

Limited capacity for contour curvature in iconic memory

We measured the difference threshold for contour curvature in iconic memory by using the cued discrimination method. The study stimulus consisting of 2 to 6 curved contours was briefly presented in the fovea, followed by two lines as cues. Subjects discriminated the curvature of two cued curves. The cue delays were 0 msec. and 300 msec. in Exps. 1 and 2, respectively, and 50 msec. before the study offset in Exp. 3. Analysis of data from Exps. 1 and 2 showed that the Weber fraction rose monotonically with the increase in set size. Clear set-size effects indicate that iconic memory has a limited capacity. Moreover, clear set-size effect in Exp. 3 indicates that perception itself has a limited capacity. Larger set-size effects in Exp. 1 than in Exp. 3 suggest that iconic memory after perceptual process has limited capacity. These properties of iconic memory at threshold level are contradictory to the traditional view that iconic memory has a high capacity both at suprathreshold and categorical levels.     

Spare the rod and spoil the icon

Short-term visual storage was investigated with a successive field paradigm, so that correct performance depended upon combining visual information from two targets that were never on simultaneously. In the first two experiments, the stimuli consisted of two slides, each containing a 10' red dot on a gray surround, and separated by an interstimulus interval (ISI) from 20 to 400 msec. Subjects had to determine if the dots were vertically or horizontally aligned. In Experiment 1, the stimuli had either no contrast for the rods or no luminance contrast for the cones or high contrast. At short ISIs the cone contrast determined performance, whereas at long ISIs the rod contrast determined performance. In fact, when the dots were invisible to the rods, the task was impossible for long ISIs. In Experiment 2, performance was compared for zero log rod contrast and for small departures from zero. Even a small departure from zero log rod contrast resulted in above-chance performance. In Experiment 3, the stimuli were luminous rectangles and the task was to decide whether or not a 4' spatial gap was present between the two successively presented rectangles. Wavelength, luminance, and ISI were varied under both photopic and scotopic adapting conditions. The result was that the rods performed the task for ISIs of 150 msec or longer under scotopic conditions and under the photopic conditions that we were able to test. The results of the experiments taken together are consistent with the hypothesis that the cone icon is short, whereas the rod icon is robust and long lasting.     

Behavioral Goals Constrain the Selection of Visual Information

Although there is agreement that attentional processes are limited, the necessary conditions for such limitation have not been determined. We investigated whether behavioral goals are sufficient to constrain the selection of visual information. In two tasks, subjects were presented with targets and distractors that varied on two dimensions (e.g., color and letter). In separate conditions, the subjects' goal was to identify only one dimension of the target while ignoring the second dimension and ignoring the distractors. In both tasks, peripheral distractors interfered with target selection only when the targets and distractors differed on the goal-relevant dimension. When the goal was changed, the pattern of interference from the same stimuli was reversed, so that distractors affected target selection only according to the new goal. These results suggest that behavioral goals constrain the selection of visual information to a greater extent than the physical characteristics of the visual information.     

Capacity limitation in short-term visual memory for contour curvature

I measured the difference threshold for contour curvature in short-term visual memory (STVM) using a two-interval forced-choice partial discrimination task. In experiments 1 and 2, the study stimulus consisting of 1 to 4 curved contours was briefly presented. It was followed by a single contour stimulus after a retention interval. The subjects judged if the test stimulus had a higher or lower curvature than the corresponding study contour. The results of experiment 1 showed that the Weber fraction increased monotonically with increasing set size. The results of experiment 2 clarified that the set-size effect was not due to a temporal limitation in encoding resulting from the short exposure time. In experiment 3, the study stimuli always consisted of 4 items, but the numbers of memorised items were different in each condition. Nevertheless, the results showed the set-size effect, which indicated that its occurrence depended largely on the capacity limitation in short-term visual memory (STVM) storage. Otherwise, the Weber fraction was not hugely higher for set size 4 compared with set size 1. It was concluded that only 1 object could be retained in STVM with high fidelity, but that at least 4 objects could be retained in STVM with some degree of fidelity.     

工作记忆的存储时间及目标时距对时间知觉的影响

本研究操作记忆信息与计时开始之间的时间间隔（ISI）和目标时距,探讨工作记忆影响时间判断的灵活性。被试首先记忆一个客体,然后在每个trial的最后判断测试刺激是否与记忆项相同;在延迟阶段,被试完成时间判断任务,即判断相继出现的两个刺激的时距哪个更长（或更短）。时间任务中的一个刺激与记忆内容完全相同,相应的另一个刺激与记忆内容在形状和颜色上都不同。重复条件下,被试忽略第一个刺激,仅完成时间判断任务。结果发现,时间间隔（ISI）短时,记忆匹配条件下的准确率更高,匹配刺激延长了知觉的时间;但随着时间间隔的增加,工作记忆匹配对时间判断的影响降低甚至消失。并且,长或短ISI,记忆任务或重复条件下,目标时距长时,记忆匹配反而缩短了知觉的时间。研究说明工作记忆对时间判断的影响是灵活的,受到注意或工作记忆等高级认知系统的调控。     

Semantic priming and identification of near threshold primes in a lexical decision task

Three experiments examined both the impact of semantic analysis of 50-msec, masked visual primes on a target response and the impact of semantic analysis of the target on a prime response. The first two experiments used a prime-target interval of 1000 msec. In Experiment 1, subjects reported the identity of each prime: (a) after a lexical decision about the target, (b) both before and after a lexical decision, or (c) after a target detection response. Prime report after both types of target response showed retroactive priming in which report was facilitated by related targets and inhibited by unrelated targets. Analyses of lexical decision latency and accuracy conditionalized on prime report showed that semantic priming was restricted to reported related primes. In Experiment 2, subjects made no overt response to the primes. Priming was conditionalized on recognition of the primes on a subsequent test. The pattern was the same as Experiment 1: There was priming only for recognized primes; recognition memory showed a pattern consistent with retroactive priming. Experiment 3 also conditionalized priming on recognition performance but used a prime-target interval of only 250 msec. Again, semantic priming was found only for recognized primes, and recognition memory revealed retroactive priming. Retroactive priming indicates an interdependency between prime and target processing that needs to be incorporated into models of semantic priming.     

Distinctiveness and serial position effects in tonal sequences.

The proportion-of-the-total-duration rule (Kidd & Watson, 1992) states that the detectability of a change in a component of a tonal sequence can be predicted by the proportional duration of the changed component relative to the length of the sequence as a whole. A similar viewpoint relies on temporal distinctiveness to account for primacy, recency, and other serial position effects in memory (Murdock, 1960; Neath, 1993a, 1993b). Such distinctiveness models predict that an item will be remembered if it is more distinctive along some dimension relative to possible competitors. Three experiments explored the relation between distinctiveness and proportional duration by examining the effects of the proportion of the total duration of a tone in a sequence, serial position, and interstimulus interval (ISI) on the detection of a change in one component of a tonal sequence. Experiment 1 replicated the basic effect with relatively untrained subjects and a fixed frequency difference. Experiment 2 showed that distinctiveness holds for tonal sequences and a same/different task. Experiment 3 combined the two to show that proportional duration, ISI, and position of the changed tone all contribute to discrimination performance. The present research combines theories that have been proposed in the psychophysics and memory fields and suggests that a comprehensive principle based on relative distinctiveness may be able to account for both perceptual and memory effects.     

Distinctiveness and serial position effects in tonal sequences

The proportion-of-the-total-duration rule (Kidd & Watson, 1992) states that the detectability of a change in a component of a tonal sequence can be predicted by the proportional duration of the changed component relative to the length of the sequence as a whole. A similar viewpoint relies on temporal distinctiveness to account for primacy, recency, and other serial position effects in memory (Murdock, 1960; Neath, 1993a, 1993b). Such distinctiveness models predict that an item will be remembered if it is more distinctive along some dimension relative to possible competitors. Three experiments explored the relation between distinctiveness and proportional duration by examining the effects of the proportion of the total duration of a tone in a sequence, serial position, and interstimulus interval (ISI) on the detection of a change in one component of a tonal sequence. Experiment 1 replicated the basic effect with relatively untrained subjects and a fixed frequency difference. Experiment 2 showed that distinctiveness holds for tonal sequences and a same/different task. Experiment 3 combined the two to show that proportional duration, ISI, and position of the changed tone all contribute to discrimination performance. The present research combines theories that have been proposed in the psychophysics and memory fields and suggests that a comprehensive principle based on relative distinctiveness may be able to account for both perceptual and memory effects.     

Working memory capacity predicts search accuracy for novel as well as repeated targets

Visual search behaviour is guided by mental representations of targets that direct attention toward relevant features in the environment. Electrophysiological data suggests these target templates are maintained by visual working memory during search for novel targets and rapidly transfer to long term memory with target repetition. If this account is correct, an individual’s working memory capacity should be more predictive of search performance for novel targets than repeated targets. Across six experiments, we tested this hypothesis using both single (Experiments 5 and 6) and multiple (Experiments 1–4) target search tasks with three different types of stimuli (real world objects, letters, and triple conjunction shapes). Each target set was repeated for six consecutive trials. In addition, we estimated visual working memory capacity using a change detection working memory task. Overall, working memory capacity did not predict response time or efficiency in the visual search task. However, working memory capacity was equally predictive of search accuracy for both novel and repeated targets. These results suggest that working memory requirements do not substantially differ between novel and repeated target search, and working memory capacity may continue to play an important role in the encoding or maintenance of target representations after they are presumed to be in long term memory.     

Hybrid search in the temporal domain: Evidence for rapid,serial logarithmic search through memory

In hybrid search, observers memorize a number of possible targets and then search for any of these in visual arrays of items. Wolfe (2012) has previously shown that the response times in hybrid search increase with the log of the memory set size. What enables this logarithmic search of memory? One possibility is a series of steps in which subsets of the memory set are compared to all items in the visual set simultaneously. In the present experiments, we presented single visual items sequentially in a rapid serial visual presentation (RSVP) display, eliminating the possibility of simultaneous testing of all items. We used a staircasing procedure to estimate the time necessary to effectively detect the target in the RSVP stream. Processing time increased in a log–linear fashion with the number of potential targets. This finding eliminates the class of models that require simultaneous comparison of some memory items to all (or many) items in the visual display. Experiment 3 showed that, similar to visual search, memory search efficiency in this paradigm is influenced by the similarity between the target set and the distractors. These results indicate that observers perform separate memory searches on each eligible item in the visual display. Moreover, it appears that memory search for one item can proceed while other items are being categorized as “eligible” or “not eligible.”     

The equivalence of target and nontarget processing in visual search

A comparison of a forced-choice visual search task with an item recognition task did not support Neisser’s (1967) hypothesis of a preattentive stage that processes targets and nontargets differentially. In the forced-choice condition, Ss indicated which of two items in a visual display was a target; in item recognition, Ss determined whether or not the single item in the visual display was a target. The size of the memorized set of possible targets was varied from one to six items for both tasks. Latencies increased linearly with memory set size in both conditions; the slopes for forced choice and item recognition were 41.8 and 27.9 msec per item, respectively. The ratio of 1.38 between the two slopes was well fit by Sternberg’s (1967) item recognition model, which predicts a ratio of 1.50.     

Searching for one versus two identical targets: when visual search has a memory

A recent study has suggested that observers' visual explorations of the external world can proceed unimpaired when the visual environment precludes the operation of memory processes (as, for instance, when the display elements change locations every 100 ms). One theoretical limitation of this study was that distractors were the only elements that had the potential to be tagged during visual search. The present study sought to clarify the amnesic-search hypothesis by investigating whether memory processes can guide search in other contexts in which targets also have the potential to be tagged. Accordingly, the experimental conditions of the previous study were repeated using a different search task in which observers had to decide whether one target or two were present among a variable number of similar distractors. Under these search conditions, the present findings provided strong evidence that memory processes can guide visual search.     

Orienting to targets by looking and pointing: Parallels and interactions in ocular and manual performance

Orienting to a target by looking and pointing is examined for parallels between the control of the two systems and interactions due to movement of the eyes and limb to the same target. Parallels appear early in orienting and may be due to common processing of spatial information for the ocular and manual systems. The eyes and limb both have shorter response latency to central visual and peripheral auditory targets. Each movement also has shorter latency and duration when the target presentation is short enough (200 msec) that no analysis of feedback of the target position is possible during the movement. Interactions appear at many stages of information processing for movement. Latency of ocular movement is much longer when the subject also points, and the eye and limb movement latencies are highly correlated for orienting to auditory targets. Final position of eyes and limb are significantly correlated only when target duration is short (200 msec). This illustrates that sensory information obtained before the movement begins is an important, but not the only, source of input about target position. Additional information that assists orienting may be passed from one system to another, since visual information gained by looking aided pointing to lights and proprioceptive information from the pointing hand seemed to assist the eyes in looking to sounds. Thus the production of this simple set of movements may be partly described by a cascade-type process of parallel analysis of spatial information for eye and hand control, but is also, later in the movement, assisted by cross-system interaction.     

Retention of time information in forced-choice duration discrimination

We explore the effect on performance in a forced-choice duration-discrimination task of varying the interstimulus interval (ISI) from 0 to 2 sec. The durations were brief empty intervals (115–285 msec) bounded by very brief auditory pulses. Performance improved as the ISI increased from 0 to 1/2 sec, but a further increase in ISI up to 2 sec resulted in little further change in performance. The “time information” derived from a brief interval bounded by auditory pulses does not appear to be susceptible to the very short-term perceptual memory loss inferred in other auditory discriminations.     

Absence tasks of the set-size effect in memory-search in the absence of a preprobe delay

In the traditional Sternberg (1966) paradigm, response latency increases linearly with increases in the size of the positive set (the set-size effect). The results of four experiments converge on the conclusion that this set-size effect depends on a delay before the presentation of the memory probe. In Experiment 1, subjects were required to respond as soon as a repetition occurred in a series of digits. Despite the similarity of this task to memory-search tasks that invariably show set-size effects, there was no increase in response latency with increasing series length. Neither the inclusion of negative trials (Experiment 2) nor the explicit designation of the test digit (Experiment 3) resulted in the typical set-size effect. However, the introduction of a 1-sec preprobe delay (Experiment 4) resulted in a set-size effect of 31 msec/item.     

Spatial cuing does not affect the magnitude of the attentional blink

Identification of the second of two targets is impaired when the second target is presented less than about 500 msec after the first. Nieuwenstein, Chun, van der Lubbe, and Hooge (2005, Experiment 4) reported that the magnitude of this attentional blink (AB) is reduced when the location of the second target is precued. Here we show how that finding resulted from an artifact brought about by a ceiling imposed by data limitation. Instead of using an accuracy measure, the present work used a dynamic threshold-tracking procedure that was not constrained by a performance ceiling. The results show that, when the ceiling is removed, spatial cuing does not affect and is not affected by the AB. These results are consistent with the hypothesis that cue localization and target identification may take place along separate (dorsal and ventral) visual pathways.     

The effect of spatial and nonspatial contextual information on visual object memory

Recent research has found visual object memory can be stored as part of a larger scene representation rather than independently of scene context. The present study examined how spatial and nonspatial contextual information modulate visual object memory. Two experiments tested participants’ visual memory by using a change detection task in which a target object's orientation was either the same as it appeared during initial viewing or changed. In addition, we examined the effect of spatial and nonspatial contextual manipulations on change detection performance. The results revealed that visual object representations can be maintained reliably after viewing arrays of objects. Moreover, change detection performance was significantly higher when either spatial or nonspatial contextual information remained the same in the test image. We concluded that while processing complex visual stimuli such as object arrays, visual object memory can be stored as part of a comprehensive scene representation, and both spatial and nonspatial contextual changes modulate visual memory retrieval and comparison.     

The right hemisphere advantage in visual change detection depends on temporal factors

What accounts for the Right Hemisphere (RH) functional superiority in visual change detection? An original task which combines one-shot and divided visual field paradigms allowed us to direct change information initially to the RH or the Left Hemisphere (LH) by deleting, respectively, an object included in the left or right half of a scene presented centrally. We manipulated the perceptual salience and semantic relevance of the change as well as the duration of the Inter-Stimulus Interval (ISI) between the scenes in order to clarify the role of the RH in memory and attention processes, and to explore whether lengthening the ISI would enhance the contribution of the LH. When analyzing data collapsed over the two levels (high vs. low) of salience and of relevance, changes were better detected in the left visual field (lvf) than in the right visual field (rvf) in the case of a short ISI, while no difference emerged in the case of a long ISI. Moreover, lengthening the ISI resulted in a performance decrement in the lvf, both for accuracy and response speed. The fact that the RH superiority was limited to short intervals indicates that stimulus-driven orienting contributes more than perceptual processing to this hemispheric asymmetry. When considering perceptual and semantic properties of the change, the effect of the ISI duration seemed to specifically emerge in the case of low relevance, with an enhancement of accuracy in the rvf when comparing the long with the short ISI. This suggests that the ISI influence on hemispheric performance operates on different levels.