Further evidence concerning scanning and sampling assumptions of visual detection models

Following extended training in a visual detection task, functions were determined for individual Ss relating latency of detection responses to number of redundant signal elements embedded in tachistoscopic displays of letters and to distance between signal elements. Latency proved invariant with respect to number of redundant signals and varied non-monotonically with distance. Of the several types of models considered in relation to previous studies, substantial support was forthcoming only for an independent stimulus sampling model. It is suggested that the detection method effects a relatively clear separation of the perceptual from the mnemonic aspects of the standard visual apprehension experiment, and that the sampling process may constitute only the first phase of a more general model which includes both parallel and serial information processing.     

How serial is serial processing in vision?

Visual search for an element defined by the conjunction of its colour and orientation has previously been shown to be a serial processing task since reaction times increase linearly with the number of distractor elements used in the display. Evidence is presented that there are parallel processing constituents to this serial search. Processing time depended on the ratio of the number of the two distractor types used, suggesting that only one type was scanned. Which type was scanned also depended on the distractor ratio, indicating that this decision was made after stimulus presentation and was based on a parallel figure-ground separation of the stimulus elements. Furthermore, in accordance with this serial scanning model, there was an increase in processing speed (elements scanned per second) with increase in number of elements to be scanned. This increased efficiency suggests that clumps of elements were processed synchronously. Under the stimulation conditions used, clumps contained six to sixteen elements and each clump was processed in 50-150 ms.     

Reaction time in relation to display size and correctness of response in forced-choice visual signal detection

Response times were recorded in a two-alternative, forced choice visual detection situation. Stimulus displays, presented tachistoscopically, were randomly selected consonant letters distributed in random subsets of cells of a matrix. Display sizes in Experiment I were 8, 12, and 16 letters; in Experiment 2–1, 4, and 8 letters; on each trial S operated a key to indicate which member of a predesignated pair of letters (signal elements) was present in a given display. Correct response times, on the average, increased uniformly with display size. Incorrect response times were uniformly greater than correct response times and, except for a reduction in the case of one element displays, were constant over display size. These relationships appear to require a modification of one assumption in the earlier proposed serial processing model for tachistoscopic perception.     

Reaction time in relation to display size and correctness of response in forced-choice visual signal detection

Response times were recorded in a two-alternative, forced choice visual detection situation. Stimulus displays, presented tachistoscopically, were randomly selected consonant letters distributed in random subsets of cells of a matrix. Display sizes in Experiment 1 were 8, 12, and 16 letters; in Experiment 2—1, 4, and 8 letters; on each trial S operated a key to indicate which member of a predesignated pair of letters (signal elements) was present in a given display. Correct response times, on the average, increased uniformly with display size. Incorrect response times were uniformly greater than correct response times and, except for a reduction in the case of one element displays, were constant over display size. These relationships appear to require a modification of one assumption in the earlier proposed serial processing model for tachistoscopic perception.     

Effects of noise similarity and redundancy on the information processed from brief visual displays

When a target letter form is presented tachistoscopicalfy and embedded in a matrix of irrelevant letter forms (noise), correct detections of the target at a given matrix size can be used to estimate indirectly the amount of information processed from the display, according to a model developed by Estes. In three experiments, the physical similarity between noise and target letters and noise-letter redundancy were varied to determine if the detection mechanism was responsive to these variables. It was found that increases in the physical differences between noise and target letters and increases in redundancy facilitated detection. It is argued that these data support a detection model that assumes parallel rather than serial information processing.     

Detecting conjunctions of color and form in parallel

Certain theories of visual attention assume that at least one processing stage must be serial when the target of search is defined as the conjunction of two or more separable features. To explain why conjunction-search response times do not always form linearly increasing functions of display size, recent versions of this general model have posited the existence of an early parallel process that guides the serial stage toward display elements that are likely targets. Other models have relaxed the seriality assumption, allowing for a limited number of parallel decisions. In the three experiments reported here, a redundant-target detection task was used with conjunctively defined targets and display sizes of two (Experiment 1), one or two (Experiment 2), and six (Experiment 3). In all three experiments, strong evidence for parallel processing was observed. The implications for models of elementary visual processes are discussed.     

Sensory stores and high-speed scanning

A model of scanning based on separate sensory and short-term stores was suggested and tested. The experiment used a probe recognition method, with set size varied from 1 to 8 and a fast presentation rate. A masking procedure was used to vary the number of items available in a possible sensory store. Although conditions were such as to maximize the chances of detecting an effect, none was found: the possible size of the sensory store had no effect on reaction time whatsoever. Other aspects of the data lent little support to a serial exhaustive scanning model, but a previously proposed parallel processing model fared better. Finally, not only was the function for positive probes steeper than that for negative probes but also there was a crossover effect as well. This crossover is not without precedent and may indicate the need for consideration of both accuracy and latency in high-speed scanning studies.     

Evidence for a parallel input serial analysis model of word processing

A parallel input serial analysis (PISA) model of word processing was developed and tested. The goal was to expand on the "critical processing duration" hypothesis of Johnson, Allen, and Strand (1989) so that both single-word and multiple-word presentation, letter detection data could be explained. In Experiments 1-3 four different word frequency categories on a single-presentation, letter detection task were used. These three experiments indicated that there was a curvilinear relationship between word frequency and letter detection reaction time (RT). That is, letter detection RTs for medium-high-frequency words were significantly longer than letter detection RTs for very-high-, low-, and very-low-frequency words. These results support the PISA model rather than the Healy, Oliver, and McNamara (1987) version of the unitization model. In Experiments 4-5 multiple-presentation (i.e., two words), letter detection tasks were used. The PISA model could also account for the results from these two experiments, but the unitization model could not.     

Serial search for targets defined by divergence or deformation of optic flow.

The optic flow field can be described in terms of the local differential measures, divergence, deformation, and rotation, which are informative about observer motion and the 3-D structure of the environment. Does an explicit representation of these measures exist in human visual processing in the form of a feature map? Triesman's criteria were used to investigate this; ie is there 'pop-out' for a target defined as different in local divergence or deformation from surrounding elements, or is a serial search necessary? The stimulus arrays contained 3, 5, or 9 square or rectangular elements, which each underwent repeated cycles of expansion, contraction, or deformation. The time required to detect a target undergoing the opposite transformation increased steeply with the number of elements, implying very slow serial search. (The mean time was 210 ms per element for divergence targets and 542 ms per element for deformation). The process was clearly still serial when the density and number of elements was increased up to 48 in an array 2.16 deg x 2.16 deg. In contrast, a single line element undergoing the opposite direction of translation motion to the rest of the display did show pop-out. It is concluded that no parallel processes seem to exist which are sensitive to the spatial uniformity of divergence and of deformation of optic flow. These differential properties may be derived as conjunctions of signals from a primary process which extracts local velocity. This result contrasts with our findings for targets defined by stereo disparity gradient, which show parallel processing in analogous experiments.     

Observers detecting a signal in two multiple observation tasks

Signal detection theory (SDT) is tested against a high threshold and a low threshold theory for two kinds of unusual tasks. In the first, the detection of the signal uses the central level of information processing; in the second, the lack of information originates in the stimulus itself rather than in the sensory organs' limitations. The observation is repeated a fixed number of times for the two tasks, and the observed integration of successive information is compared with predictions derived from the SDT and from threshold theories. All results strongly support the SDT which can thus be considered as a general model for risky binary decision-making insofar as these experiments are representative. This would be an argument to replace the classical analysis of the performance in terms of time and number of errors by a new analysis in terms of capability and strategy. The data show how the subjects are more selective for detecting the signal when the task is more difficult, and at the beginning and the end of a series of presentations. The organization of the decision-criteria of the rating-procedure varies with the difficulty of the condition, and with the number of observations in order to allow the subject to discriminate between likelihood ratios which become more precise.     

The statistical analysis of concurrent detection ratings

The concurrent detection task is a powerful method for assessing interactions in the processing of two sensory signals. On each trial, a stimulus is presented that is composed of one, both, or neither signal, and the observer makes a detection rating for each stimulus. A classical bivariate signal-detection analysis applies to these data, but is limited by its inability to differentiate certain types of sensory interactions from more cognitive components, and by the lack of an associated testing procedure. The present paper presents an alternative analysis, based on the contingency table of sensory ratings. Six classes of effect can be distinguished and tested: (1) simple response bias, (2) detection of the two signals, (3) interference of each signal on the response to the other signal, (4) sensory and response correlation, (5) bivariate response biases, and (6) higher order association. Complete computational detail is provided.     

Parallel processing in visual search asymmetry

The difficulty of visual search may depend on assignment of the same visual elements as targets and distractors-search asymmetry. Easy C-in-O searches and difficult O-in-C searches are often associated with parallel and serial search, respectively. Here, the time course of visual search was measured for both tasks with speed-accuracy methods. The time courses of the 2 tasks were similar and independent of display size. New probabilistic parallel and serial search models and sophisticated-guessing variants made predictions about time course and accuracy of visual search. The probabilistic parallel model provided an excellent account of the data, but the serial model did not. Asymptotic search accuracies and display size effects were consistent with a signal-detection analysis, with lower variance encoding of Cs than Os. In the absence of eye movements, asymmetric visual search, long considered an example of serial deployment of covert attention, is qualitatively and quantitatively consistent with parallel search processes.     

Some evidence for a limited capacity parallel selfterminating process in simple visual search tasks.

An experiment was carried out in order to investigate the questions of exhaustive vs selfterminating and serial vs parallel processing in a simple visual search task. In the experiment, 1, 2 or 3 letters were placed on an imaginary circle round the fixation point. Two different letters were used, one of which was defined as the ‘signal’. Ss had to respond ‘yes’ when one or more signals were in the display, ‘no’ otherwise. The number of signals in the display was varied from ‘no signals’ to ‘all signals’.A decrease in latencies with increasing number of signals for a fixed number of elements presented was observed indicating a selfterminating search. A decrease in latencies with increasing number of elements when only signals were presented was taken as evidence for a parallel selfterminating process. A further analysis of the error data showed that a limited capacity system had to be assumed. It was shown that it was possible to construct the overall pattern of latencies over conditions from the error data obtained.     

Visual detection in relation to display size and redundancy of critical elements I

Visual detection was studied in relation to displays of discrete elements, randomly selected consonant letters, distributed in random subsets of cells of a matrix, the S being required on each trial to indicate only which member of a predesignated pair of critical elements was present in a given display. Experimental variables were number of elements per display and number of redundant critical elements per display. Estimates of the number of elements effectively processed by a S during a 50 ms. exposure increased with display size, but not in the manner that would be expected if the S sampled a fixed proportion of the elements present in a display of given area. Test-retest data indicated substantial correlations over long intervals of time in the particular elements sampled by a S from a particular display. Efficiencies of detection with redundant critical elements were very close to those expected on the hypothesis of constant sample size over trials for any given display size and were relatively invariant with respect to distance between critical elements.     

Mechanisms of visual threat detection in specific phobia

People with anxiety or stress-related disorders attend differently to threat-relevant compared with non-threat stimuli, yet the temporal mechanisms of differential allocation of attention are not well understood. We investigated two independent mechanisms of temporal processing of visual threat by comparing spider-phobic and non-fearful participants using a rapid serial visual presentation task. Consistent with prior literature, spider phobics, but not non-fearful controls, displayed threat-specific facilitated detection of spider stimuli relative to negative stimuli and neutral stimuli. Further, signal detection analyses revealed that facilitated threat detection in spider-phobic participants was driven by greater sensitivity to threat stimulus features and a trend towards a lower threshold for detecting spider stimuli. However, phobic participants did not display reliably slowed temporal disengagement from threat-relevant stimuli. These findings advance our understanding of threat feature processing that might contribute to the onset and maintenance of symptoms in specific phobia and disorders that involve visual threat information more generally.     

Detection and recognition of intensity changes in tone and noise: The detection-recognition disparity

Recognition of increments vs decrements in auditory intensity improves with signal duration, relative to detection of increments and decrements. This effect obtains whether the background stimulus is a tone, noise, or a tone with noise masker, and is largely uninfluenced by the rise time of the signals. These data are inconsistent with detection models in which the O makes only one sensory observation during each observation interval, but can be described in terms of a neural timing model in which transients play a critical role.     

The involvement of central attention in visual search is determined by task demands

Attention, the mechanism by which a subset of sensory inputs is prioritized over others, operates at multiple processing stages. Specifically, attention enhances weak sensory signal at the perceptual stage, while it serves to select appropriate responses or consolidate sensory representations into short-term memory at the central stage. This study investigated the independence and interaction between perceptual and central attention. To do so, I used a dual-task paradigm, pairing a four-alternative choice task with a visual search task. The results showed that central attention for response selection was engaged in perceptual processing for visual search when the number of search items increased, thereby increasing the demand for serial allocation of focal attention. By contrast, central attention and perceptual attention remained independent as far as the demand for serial shifting of focal attention remained constant; decreasing stimulus contrast or increasing the set size of a parallel search did not evoke the involvement of central attention in visual search. These results suggest that the nature of concurrent visual search process plays a crucial role in the functional interaction between two different types of attention.     

Multimodal signal detection: Independent decisions vs. integration

The pooling of information from simultaneous, spatially congruent auditory and visual stimuli is examined in a signal detection task. The paradigm used permits discrimination among a number of models of the decision mechanisms involved in processing multiple component stimuli. Parameter-free predictions are presented for the weighted integration model and for three versions of the independent decisions model. The data support an independent decisions model of the bimodal detection process in which attention is shared equally between modalities.     

The Lower Visual Search Efficiency For Conjunctions Is Due To Noise and Not Serial Attentional Processing

Models of human visual processing start with an initial stage with parallel independent processing of different physical attributes or features (e.g. color, orientation, motion). A second stage in these models is a temporally serial mechanism (visual attention) that combines or binds information across feature dimensions. Evidence for this serial mechanism is based on experimental results for visual search. I conducted a study of visual search accuracy that carefully controlled for low-level effects: physical similarity of target and distractor, element eccentricity, and eye movements. The larger set-size effects in visual search accuracy for briefly flashed conjunction displays, compared with feature displays, are quantitatively predicted by a simple model in which each feature dimension is processed independently with inherent neural noise and information is combined linearly across feature dimensions. The data are not predicted by a temporally serial mechanism or by a hybrid model with temporally serial and noisy processing. The results do not support the idea that a temporally serial mechanism, visual attention, binds information across feature dimensions and show that the conjunction-feature dichotomy is due to the noisy independent processing of features in the human visual system.     

A signal detection model predicts the effects of set size on visual search accuracy for feature, conjunction, triple conjunction, and disjunction displays

Recently, quantitative models based on signal detection theory have been successfully applied to the prediction of human accuracy in visual search for a target that differs from distractors along a single attribute (feature search). The present paper extends these models for visual search accuracy to multidimensional search displays in which the target differs from the distractors along more than one feature dimension (conjunction, disjunction, and triple conjunction displays). The model assumes that each element in the display elicits a noisy representation for each of the relevant feature dimensions. The observer combines the representations across feature dimensions to obtain a single decision variable, and the stimulus with the maximum value determines the response. The model accurately predicts human experimental data on visual search accuracy in conjunctions and disjunctions of contrast and orientation. The model accounts for performance degradation without resorting to a limited-capacity spatially localized and temporally serial mechanism by which to bind information across feature dimensions.