Serial processing in reading aloud: no challenge for a parallel model

K. Rastle and M. Coltheart (1999) challenged parallel models of reading by showing that the cost of irregularity in low-frequency exception words was modulated by the position of the irregularity in the word. This position-of-irregularity effect was taken as strong evidence of serial processing in reading. This article refutes Rastle and Coltheart's theoretical conclusions in 3 ways: First, a parallel model, the connectionist dual process model (M. Zorzi, G. Houghton, & B. Butterworth, 1998b), produces a position-of-irregularity effect. Second, the supposed serial effect can be reduced to a position-specific grapheme-phoneme consistency effect. Third, the position-of-irregularity effect vanishes when the experimental data are reanalyzed using grapheme-phoneme consistency as the covariate. This demonstration has broader implications for studies aiming at adjudicating between models: Strong inferences should be avoided until the computational models are actually tested.     

Serial and within-stage independent parallel model equivalence on the minimum completion time

A set of functional equations is investigated that, when satisfied, yields equivalence on the minimum completion time between serial models and parallel models having the property of within-stage independence. Within these classes of models, it is shown that any parallel model is equivalent to a serial model, but not any serial model is equivalent to a parallel model. The necessary and sufficient conditions for a serial model to have an equivalent parallel counterpart are observed and then two sufficient conditions on the survivor functions that produce this result are exhibited. A number of examples satisfying the various theorems are discussed and a special case leading to an extension of a property of exponential distributions is derived.     

Top-down inhibition of search distractors in parallel visual search

In three experiments, we examined distractor inhibition in parallel ("pop-out") visual search. Distractor inhibition was measured in terms of reaction time (RT) to a simple luminance increment probe presented, after the search task response, at display locations that either contained a search distractor (on-probe) or were blank (off-probe). When the search stimuli remained in view, the on-probe (relative to off-probe) RT cost was larger than in a baseline condition in which observers had only to passively view, rather than search, the display. This differential on-probe RT cost, which discounts effects of masking, was interpreted as a measure of distractor inhibition associated with target selection in parallel visual search. Taken together, the results argue that the distractor inhibition is an object-based and local phenomenon that affects all distractors (of a particular type) in an equal manner.     

The role of crowding in parallel search: Peripheral pooling is not responsible for logarithmic efficiency in parallel search

Recent results from our laboratory showed that, in fixed-target parallel search tasks, reaction times increase in a logarithmic fashion with set size, and the slope of this logarithmic function is modulated by lure-target similarity. These results were interpreted as being consistent with a processing architecture where early vision (stage one) processes elements in the display in exhaustive fashion with unlimited capacity and with a limitation in resolution. Here, we evaluate the contribution of crowding to our recent logarithmic search slope findings, considering the possibility that peripheral pooling of features (as observed in crowding) may be responsible for logarithmic efficiency. Factors known to affect the strength of crowding were varied, specifically: item spacing and similarity. The results from three experiments converge on the same pattern of results: reaction times increased logarithmically with set size and were modulated by lure-target similarity even when crowding was minimized within displays through an inter-item spacing manipulation. Furthermore, we found logarithmic search efficiencies were overall improved in displays where crowding was minimized compared to displays where crowding was possible. The findings from these three experiments suggest logarithmic efficiency in efficient search is not the result peripheral pooling of features. That said, the presence of crowding does tend to reduce search efficiency, even in “pop-out” search situations.     

Preattentive perception of multiple illusory line-motion: a formal model of parallel independent-detection in visual search

The phenomenon referred to as illusory line-motion (ILM; O. Hikosaka, S. Miyauchi, & S. Shimojo, 1993a) has been described as a measure of the local facilitation of attention gradient. However, J. Kawahara, K. Yokosawa, S. Nishida, and T. Sato (1996) have demonstrated a spatially parallel search for an "odd man out" in the ILM direction. Apart from showing preattentive ILM perception in terms of an analogy between line-motion and apparent motion, the authors examined whether ILM perception is possible without attention from another point of view. Four experiments revealed that the ILM target can be detected in parallel without invoking attentional facilitation and invalidated the possible contribution of attentional set in parallel ILM search. Participants were able to correctly detect the ILM target among multiple nontargets, even when the line orientation was changed from trial to trial. The authors' independent-detection model predicted ILM search performance well on several occasions. These findings strongly support a preattentive and stimulus-driven explanation of ILM perception.     

Serial deployment of attention during visual search

This study examined whether objects are attended in serial or in parallel during a demanding visual search task. A component of the event-related potential waveform, the N2pc wave, was used as a continuous measure of the allocation of attention to possible targets in the search arrays. Experiment 1 demonstrated that the relative allocation of attention shifts rapidly, favoring one item and then another. In Experiment 2, a paradigm was used that made it possible to track the absolute allocation of attention to individual items. This experiment showed that attention was allocated to one object for 100-150 ms before attention began to be allocated to the next object. These findings support models of attention that posit serial processing in demanding visual search tasks.     

Serial and parallel processes in working memory after practice

Six young adults practiced for 36 sessions on a working-memory updating task in which 2 digits and 2 spatial positions were continuously updated. Participants either did 1 updating operation at a time, or attempted 1 numerical and 1 spatial operation at the same time. In contrast to previous research using the same paradigm with a single digit and a single dot, dual-task costs were not eliminated with practice. Costs of switching between digits and between spatial positions were found throughout practice, supporting the existence of a focus of attention in working memory that can hold 1 digit and 1 spatial position simultaneously, but is not expanded to hold 2 elements of the same kind. The results can be understood by assuming that observed limits on parallel processing, as well as on the capacity of the focus of attention, arise not from structural constraints but rather reflect the optimal configuration of the cognitive system for avoiding information cross-talk in a given task.     

Serial search and comparison of features of imagined and perceived objects

Four experiments studied the spatial information processing involved in making a series of same-different comparisons of features of two objects. When the path between successively compared features on one object was antiparallel with the corresponding path on the other, comparison of a series of features took longer and produced many more errors. These results were observed both when the objects were externally presented and when one object was imagined and the other externally presented. Knowing the location of the next feature seems much more important for effective search than does monitoring the location of the feature used in the preceding comparison. When paths between corresponding features are parallel, search of features of one object may guide search of the other. When the directions between corresponding features are incongruous, search for the next angle may produce a competition for processes or processing resources, or may produce interfering cross-talk between the spatial information processing of the concurrent search tasks. Because of incongruity, as demonstrated in this study, serial search of objects at different orientations is difficult.     

"Serial" effects in parallel models of reading

There is now considerable evidence showing that the time to read a word out loud is influenced by an interaction between orthographic length and lexicality. Given that length effects are interpreted by advocates of dual-route models as evidence of serial processing this would seem to pose a serious challenge to models of single word reading which postulate a common parallel processing mechanism for reading both words and nonwords (Coltheart, Rastle, Perry, Langdon, & Ziegler, 2001; Rastle, Havelka, Wydell, Coltheart, & Besner, 2009). However, an alternative explanation of these data is that visual processes outside the scope of existing parallel models are responsible for generating the word-length related phenomena (Seidenberg & Plaut, 1998). Here we demonstrate that a parallel model of single word reading can account for the differential word-length effects found in the naming latencies of words and nonwords, provided that it includes a mapping from visual to orthographic representations, and that the nature of those orthographic representations are not preconstrained. The model can also simulate other supposedly "serial" effects. The overall findings were consistent with the view that visual processing contributes substantially to the word-length effects in normal reading and provided evidence to support the single-route theory which assumes words and nonwords are processed in parallel by a common mechanism.     

Beyond accuracy: Epistemic flaws with statistical generalizations

What, if anything, is epistemically wrong with beliefs involving accurate statistical generalizations about demographic groups? This paper argues that there is a perfectly general, underappreciated epistemic flaw which affects both ethically charged and uncharged statistical generalizations. Though common to both, this flaw can also explain why demographic statistical generalizations give rise to the concerns they do. To identify this flaw, we need to distinguish between the accuracy and the projectability of statistical beliefs. Statistical beliefs are accompanied by an implicit representation of the statistic's modal profile. Their modal profile determines the circumstances in which they can legitimately be projected to unobserved instances. Errors in that implicit content can be compatible with the accuracy of the “bare” statistic, whilst systematically leading to downstream errors in reasoning, in a manner which reveals an epistemic flaw with an important aspect of the belief state itself.     

From parallel to serial processing: a computational study of visual search

A novel computational model of a preattentive system performing visual search is presented. The model processes displays of lines, reproduced from Wolfe, Friedman-Hill, Stewart, and O'Connell's (1992) and Treisman and Sato's (1990) visual-search experiments. The response times measured in these experiments suggest that some of the displays are searched serially, whereas others are scanned in parallel. Our neural network model operates in two phases. First, the visual displays are compressed via standard methods (principal component analysis), to overcome assumed biological capacity limitations. Second, the compressed representations are further processed to identify a target in the display. The model succeeds in fast detection of targets in experimentally labeled parallel displays, but fails with serial ones. Analysis of the compressed internal representations reveals that compressed parallel displays contain global information that enables instantaneous target detection. However, in representations of serial displays, this global information is obscure, and hence, a target detection system should resort to a serial, attentional scan of local features across the display. Our analysis provides a numerical criterion that is strongly correlated with the experimental response time slopes and enables us to reformulate Duncan and Humphreys's (1989) search surface, using precise quantitative measures. Our findings provide further insight into the important debate concerning the dichotomous versus continuous views of parallel/serial visual search.     

Serial processing of visual spatial patterns in a search paradigm

Previous experiments in this laboratory employing a search paradigm have found highly significant differences in the detectability of a briefly exposed target pattern as a function of the spatial location of the target when it is presented simultaneously with a number of discriminably different nontarget patterns. These detectability differences, at loci equidistant from the fovea, could not be accounted for by any known variation in retinal spatial resolution or by differential lateral masking effects of the target by nearby nontarget patterns. These observations led to the hypothesis that the target in these experiments was detected by a serial mechanism which "scanned" a persisting but rapidly degrading neural representation of the visual scene with increasing detection failures the later in time the scan processed the location occupied by the target. If this hypothesis is correct, then target detectability should vary inversely with the number of stimuli which must be examined. The present experiment confirmed this expectation. A mathematical model of such a serial scanning process also predicts other, less obvious, effects on target detectability which were observed when the number of nontarget patterns was changed.     

Serial attention mechanisms in visual search: A critical look at the evidence

Until a few years ago, visual search tasks were of exclusive pertinence to psychophysicists and cognitive psychologists trying to understand the operating principles and computational constraints of visual perception and visual selective attention. In recent years, cognitive neuroscience, with its powerful tools, has started to explore more directly the neuronal mechanisms underlying search performance in humans and macaques, aiming at the same general goals. New observations from a number of cognitive neuroscience approaches are promising a near future of great excitement in this field of research. This article offers a critical review of some of these recent contributions and highlights some of the interpretational problems that they pose. Received: 20 January 1998 / Accepted: 25 September 1998     

Modeling the role of parallel processing in visual search

Treisman's Feature Integration Theory and Julesz's Texton Theory explain many aspects of visual search. However, these theories require that parallel processing mechanisms not be used in many visual searches for which they would be useful, and they imply that visual processing should be much slower than it is. Most importantly, they cannot account for recent data showing that some subjects can perform some conjunction searches very efficiently. Feature Integration Theory can be modified so that it accounts for these data and helps to answer these questions. In this new theory, which we call Guided Search, the parallel stage guides the serial stage as it chooses display elements to process. A computer simulation of Guided Search produces the same general patterns as human subjects in a number of different types of visual search.     

Visual search time based on stochastic serial and parallel processings

Two stochastic latency models were developed in order to account for random fluctuations of the search time that the S consumed when he was asked to search for a single target number or one of two (or three) target numbers through the sets of small two-digit numbers randomly distributed on a white square. It was shown that the search time distribution for a single target number can be adequately described in terms of McGill’s two-step serial processing model. The parallel processing model whose structure is exponential was found to fit approximately the search time distributions for the first target number that the S detects among the two (or three) target numbers predesignated. This model, however, does not satisfactorily handle the long search times.     

Electrophysiological evidence for parallel and serial processing during visual search

Event-related potentials were recorded from young adults during a visual search task in order to evaluate parallel and serial models of visual processing in the context of Treisman's feature integration theory. Parallel and serial search strategies were produced by the use of feature-present and feature-absent targets, respectively. In the feature-absent condition, the slopes of the functions relating reaction time and latency of the P3 component to set size were essentially identical, indicating that the longer reaction times observed for larger set sizes can be accounted for solely by changes in stimulus identification and classification time, rather than changes in post-perceptual processing stages. In addition, the amplitude of the P3 wave on target-present trials in this condition increased with set size and was greater when the preceding trial contained a target, whereas P3 activity was minimal on target-absent trials. These effects are consistent with the serial self-terminating search model and appear to contradict parallel processing accounts of attention-demanding visual search performance, at least for a subset of search paradigms. Differences in ERP scalp distributions further suggested that different physiological processes are utilized for the detection of feature presence and absence.