Prelexical phonologic computation in a deep orthography: Evidence from backward masking in Hebrew

Prelexical phonological computation of print was examined in Hebrew orthography using backward masking. Target words were masked by homophonic, phonologically similar, or phonologically dissimilar nonwords. Phonological similarity between masks and targets affected detection in a nonlinear fashion: No differences were found between homophonic and phonologically similar masks, whereas phonologically dissimilar masks hindered identification dramatically. This suggests that representations computed in brief exposures are coarse grained and not detailed enough to capture fine phonetic differences.     

Automatic and pre-lexical computation of phonology in visual word identification

Abstract

Two experiments were conducted to evaluate the hypothesis that identification of a visually presented word involves phonological information that is activated pre-lexically and automatically. A backward masking paradigm was used in which a word target was followed by a pseudoword mask, followed in turn by a non-linguistic pattern mask. The stimulus materials were Serbo-Croatian. The pseudoword mask could share all but one phoneme in common with the target, or none; moreover, it could be printed in the same alphabet as the target (e.g. both stimuli printed in Cyrillic), or in the other alphabet (e.g. target in Cyrillic, mask in Roman). Word targets were always lower case, and pseudoword masks were always upper case. It was assumed that where a mask shares phonological information with the target it can compensate for the interruption in processing by continuing the activation of the phoneme units activated by the target. Such an effect would be pre-lexical because the phoneme units activated by the mask would be those activated previously during the incomplete processing of the target. Both experiments, using different onset asynchronies among the stimuli, found significantly higher levels of target identification for homophonous masking than for non-homophonous masking, in agreement with similar studies using English materials. It was also shown that alphabet congruity affected the magnitude of the phonological effect in a direction that supported an hypothesis of inhibition of the letter processing units of one alphabet by the unique letters of the other alphabet. The results were discussed in terms of phonology's role in mediating lexical access in Serbo-Croatian and English. and in terms of a network model of visual word identification in Serbo-Croatian.     

Orthographic and phonological computation in visual word recognition: Evidence from backward masking in Hebrew

We employed the backward masking paradigm to map the early computation involved in the perception of printed words in Hebrew. By independently manipulating orthographic and phonological overlap between targets and masks, we examined whether orthographic and phonological structure imposes late or early constraints on lexical access. The results demonstrated that the probability of reinstating a masked target depended on both the orthographic and the phonological overlap between the target and the mask. However, whereas phonemic structure seems to constrain lexical access only at short exposure durations, orthographic overlap also exerts its influence at longer durations.     

Testing quantitative models of backward masking

We analyzed the relationship between U-shaped and monotonic-shaped masking functions, using both computer simulations of quantitative models and experimental data. Our analysis revealed that quantitative models of backward masking predict that U-shaped masking functions should appear for weak masks and monotonic masking functions should appear for strong masks. The models predict, moreover, that for a fixed target and experimental task, as the mask changes it is possible to go from U-shaped to monotonic-shaped masking functions. Significantly, the models predict that at each stimulus onset asynchrony between the target and the mask, the U-shaped function must have weaker masking than the monotonic-shaped function. Contrary to the predictions of the models, we show an experimental situation that generates masking functions that violate this prediction.     

Phonological codes are early sources of constraint in visual semantic categorization

Two experiments were directed at early phonological activation in the semantic categorization task. In Experiment 1, briefly exposed targets homophonic to category exemplars (rows for the categorya flower), and their graphemic controls (robs), were judged for category membership with and without a backward pattern mask False positives were greater for rows than robs to the same degree under both unmasked and masked conditions. In Experiment 2, false positives were examined in the semantic categorization task under backward dichoptic masking by pseudowords that were, in turn, masked monoptically by a pattern mask. Briefly exposed homophones (e.g.,weak), masked by a phonologically similar pseudoword (“feek”), a graphemic control (“felk”), or an unrelated pseudoword (“furt”), were categorized as category exemplars (a unit of time). The difference in false positives was significant forweak-feek versusweak-furt, but not forweak-felk versusweak-furt. It was suggested that the persistence of the homophonic effects under the pattern masking of Experiment 1 and their amplification under the phonological masking of Experiment 2 were because phonological codes cohere rapidly and provide, thereby, immediately available constraints on semantic processing.     

OBJECT SUBSTITUTION:

Abstract —Can four dots that surround, but do not touch, a target shape act as a mask to reduce target discriminability? Although existing theories of metacontrast and pattern masking say "no." we report this occurs when targets appear in unpredictable locations. In three experiments, a four-dot mask was compared with a standard metacontrast mask that surrounded the target. Although accuracy was predictably different for the two masks at a central display location in Experiment I. both masks had similar strong effects on accuracy in parafoveal locations. Experiment 2 revealed that both four-dot and metacontrast masking were insensitive to contour proximity in parafoveal display locations, and Experiment 3 showed that four-dot masking could occur even at a central location if attention was distributed among several targets. We propose that targets in unattended locations are coded with low spotiotemporal resolution, leaving them vulnerable to substitution by the four dots when attention is directed to them.     

Visual masking plays two roles in the attentional blink

When two targets are displayed in rapid visual sequence and masked by trailing patterns, identification accuracy is nearly perfect for the first target but follows a U-shaped pattern over temporal lag for the second target. Three experiments examined the role of visual masking in this attentional blink. Experiment 1 compared integration and interruption masks for both targets. Although either mask was effective in producing the blink when applied to the first target, only the interruption mask was effective when applied to the second target. Experiment 2 showed that integration masking of the second target was ineffective over a wide range of accuracy levels. Combining the two forms of masking in Experiment 3 confirmed the dissociation: A combined mask and only a main effect on accuracy for the first target, whereas it produced a qualitatively different pattern over temporal lag for the second target. These results suggest that representations of the target are substituted in consciousness by that of the interruption mask when visual attention is preoccupied.     

Differential backward masking of words and letters by masks of varying orthographic structure

The effects of structural relationships between targets and masks were investigated using a backward-masking paradigm. Specifically, the masking of single letters, common fiveletter words, and five-letter pseudowords masked by a blank flash, strings of overlapped letters, pseudowords, and words was investigated. Target duration was varied from 2 to 32 msec, with mask duration held constant at 25 msec. The dependent measure was the critical interstimulus interval for correct target identification. Letters were more effectively masked than words and pseudowords. A blank mask caused the least amount of masking, followed by the overlapped letter strings, and then the word and pseudoword masks. In addition to the overall greater masking effectiveness for the three patterned masks, overlapped letter strings masked letters more effectively than they did words. The implications of current theories of masking for these results and the implications of these results for theories of word recognition were discussed.     

Asymmetric object substitution masking

Object substitution masking occurs when a lateral mask persists beyond the duration of a target, reflecting reentrant processes in vision (V. Di Lollo, J. Enns, & R. Rensink, 2000). The authors studied whether substitution masking is location specific and whether it is symmetric around the target. A brief circular display of letters was presented along with a mask that designated the target. The mask was centered on the target or 1.1 degrees to the central or to the peripheral side. Substitution masking was found even when the target and the mask were at different locations. It was asymmetric and stronger when the mask was to the peripheral side of the target than to the central side. Asymmetric substitution was observed using various masks. It could not be explained by retina acuity gradients and was not attenuated by focused attention. The authors propose that target selection triggers an asymmetric inhibitory surround that is stronger toward the central side of the target.     

Variations in backward masking with different masking stimuli: II. The effects of spatially quantised masks in the light of local contour interaction, interchannel inhibition, perceptual retouch, and substitution theories

In part I we showed that with spatially non-overlapping targets and masks both local metacontrast-like interactions and attentional processes are involved in backward masking. In this second part we extend the strategy of varying the contents of masks to pattern masking where targets and masks overlap in space, in order to compare different masking theories. Images of human faces were backward-masked by three types of spatially quantised masks (the same faces as targets, faces different from targets, and Gaussian noise with power spectra typical for faces). Configural characteristics, rather than the spectral content of the mask, predicted the extent of masking at relatively long stimulus onset asynchronies (SOAs). This poses difficulties for the theory of transient-on-sustained inhibition as the principal mechanism of masking and also for local contour interaction being a decisive factor in pattern masking. The scale of quantisation had no effect on the masking capacity of noise masks and a strong effect on the capacity of different-face masks. Also, the decrease of configural masking with an increase in the coarseness of the quantisation of the mask highlights ambiguities inherent in the re-entrance-based substitution theory of masking. Different masking theories cannot solve the problems of masking separately. They should be combined in order to create a complex, yet comprehensible mode of interaction for the different mechanisms involved in visual backward masking.     

Target-mask shape congruence impacts the type of metacontrast masking

Visual metacontrast masking may depend on the time intervals between target and mask in two qualitatively different ways: in type-A masking the smaller the mask delay from target the stronger the masking while in type-B masking maximal masking effect is obtained with a larger temporal delay of the mask. Variability in the qualitative apperance of masking functions has been explained by variability in stimuli parameters and tasks. Recent research on metacontrast masking has surprisingly shown that both of these types of functions can be found with an identical range of stimulation parameters depending on individual differences between observers. Here we show that obtaining clear-cut type-A masking depends on whether target and mask shapes are congruent or incongruent and whether observers use the cues available due to the congruence factor. Conspicuously expressed type-A masking is selectively associated with incongruent target-mask pairings. In the latter conditions target identification level significantly drops with the shortest target-to-mask delays.     

A comparison of masking by visual and transcranial magnetic stimulation: implications for the study of conscious and unconscious visual processing

Visual stimuli as well as transcranial magnetic stimulation (TMS) can be used: (1) to suppress the visibility of a target and (2) to recover the visibility of a target that has been suppressed by another mask. Both types of stimulation thus provide useful methods for studying the microgenesis of object perception. We first review evidence of similarities between the processes by which a TMS mask and a visual mask can either suppress the visibility of targets or recover such suppressed visibility. However, we then also point out a significant difference that has important implications for the study of the time course of unconscious and conscious visual information processing and for theoretical accounts of the processes involved. We present evidence and arguments showing: (a) that visual masking techniques, by revealing more detailed aspects of target masking and target recovery, support a theoretical approach to visual masking and visual perception that must take into account activities in two separate neural channels or processing streams and, as a corollary, (b) that at the current stage of methodological sophistication visual masks, by acting in more highly specifiable ways on these pathways, provide information about the microgenesis of form perception not available with TMS masks.     

Masked-target recovery requires focused attention on the target object

Flashing a homogeneous light mask after the presentation of a masked target reduces the deleterious effects of the mask, a phenomenon often called target recovery. Target recovery has been studied using masking paradigms in which a target object is presented in isolation prior to the presentation of a mask, thus capturing attention. In the present study, we examined whether target recovery is possible when a target does not benefit from attentional capture. We hypothesized that target recovery would be eliminated when a target must compete with distractors for perceptual attention. Replicating classic studies, we observed target recovery when pattern and light masks followed an isolated target. However, target recovery was not observed when a light mask followed a masked visual search target. Furthermore, using an attentional-capture paradigm we found that sudden onset search targets were recoverable whereas nononset targets were not. The present findings indicate that attentional capture by a target prior to masking plays a critical role in the subsequent recovery of the target.     

Reaction time measures of backward masking

We employed both simple and choice reaction time (RT) paradigms in which the subjects were required to respond to 3.0 cycles per degree (c/d) square-wave gratings presented to one eye, while checkerboard masks were presented at various stimulus-onset asynchronies to the other eye. No masking was evident using the simple RT paradigm, but with the choice RT task, checkerboard masks presented to the contralateral eye of three subjects resulted in substantial decreases in response speed when the test preceded the mask by stimulus-onset asynchronies of 25 to 75 ms. Masks that contained lower fundamental spatial frequencies (1.0 c/d) than the target were more effective than masks containing fundamental spatial frequencies (6.0 c/d) higher than the target, while masks that contained fundamental components identical to those in the target (3.0 c/d) produced maximum masking. The results offer support for the sustained-transient theory of visual processing and validate RT as a technique for examining spatio-temporal factors in masking.     

Vibrotactile frequency recognition: forward and backward masking effects

Forward and backward vibrotactile recognition masking was investigated in 4 subjects with 240-Hz and 160-Hz targets of 20 ms duration and four 200-Hz masks, using interstimulus intervals (ISIs) ranging from -500 to 500 ms. Two of the masks (short) were 20 ms and two (long) were 200 ms in duration. One of each set of masks was matched in subjective intensity to the targets, but the others were more intense. The range of ISIs over which masking was obtained was comparable to that found by Massaro (1970) with auditory stimuli. Both short masks produced more masking than either long mask except at short ISIs. Larger mask intensities increased masking only at very short ISIs, and longer mask durations increased backward but not forward masking.     

Perception and masking of wholes and parts

These exerpiments show that the effects of masking on reports of target lines depend on the context in which the target lines appear. Subjects viewed brief presentations of target lines either alone or in drawings of three-dimensional objects, and each target display was preceded and followed by one of several different mask stimuli. There were two main findings: (a) A mask containing a haphazard array of lines interfered more with single lines than it did with lines in objects. (b) A mask containing drawings of the object displays interfered more with lines in objects than did either of two control masks containing relatively flat, less coherent patterns. In a control condition, the object mask interfered slightly less with reports of single lines than either of the control masks did. The discussion considers how the effects obtained here bear on models of the processing of wholistic stimuli and their component parts.     

On the labile memory buffer in the attentional blink: masking the T2 representation by onset transients mediates the AB

Report of a second target (T2) is impaired when presented within 500 ms of the first (T1). This attentional blink (AB) is known to cause a delay in T2 processing during which T2 must be stored in a labile memory buffer. We explored the buffer's characteristics using different types of masks after T2. These characteristics were inferred by determining what attributes of T2 are hindered by a given form of masking. In Experiments 1-3, trailing metacontrast and four-dot masks produced ABs of equal magnitudes, implicating the onset-transient triggered by the mask as the mechanism underlying the AB and strongly suggesting a locus in early vision. In Experiment 4, metacontrast and four-dot masks were presented in a common-onset masking (COM) paradigm in which a brief, combined display of T2 and the mask was followed by a longer display of the mask alone. COM is thought to occur late in the sequence of processing events. No AB occurred with COM, confirming the critical role of the mask's onset transients and ruling out a high-level locus for the labile memory buffer.     

Practice effects in backward masking

In two experiments we demonstrate that much larger practice effects occur in a backward masking paradigm where patterned masks are used than in similar visual processing paradigms, such as lateral masking and whole report. In additional experiments we examine four possible explanations for the large practice effects: increased familiarity with the paradigm in general, learning about the targets, learning about the masks, and enhanced sensory processing. Because of failure to observe similar practice effects in related paradigms not involving backward masking and because of the sustained nature of the improvement, we reject the first explanation as a source of practice effect. Experiment 3 allowed us to reject target learning as a source of improvement as well; target sets were switched at the end of training, but no decrement in performance was observed. In Experiment 4, mask sets were switched at the end of training, revealing a significant decrement in performance. Learning about the specific masks, then, does contribute to the observed improvement. However, it is responsible for only about one third of the overall improvement in performance. The final experiment provides evidence that the residual improvement is due to enhanced sensory processing. In that experiment, training on backward masking led to a lowered threshold in a two-flash paradigm but not to a significant change in whole-report performance.     

Forward masking of faces by spatially quantized random and structured masks: On the roles of wholistic configuration, local features, and spatial-frequency spectra in perceptual identification

The forward masking of faces by spatially quantized masking images was studied. Masks were used in order to exert different types of degrading effects on the early representations in facial information processing. Three types of source images for masks were used: Same-face images (with regard to targets), different-face images, and random Gaussian noise that was spectrally similar to facial images. They were all spatially quantized over the same range of quantization values. Same-face masks had virtually no masking effect at any of the quantization values. Different-face masks had strong masking effects only with fine-scale quantization, but led to the same efficiency of recognition as in the same-face mask condition with the coarsest quantization. Moreover, compared with the noise-mask condition, coarsely quantized different-face masks led to a relatively facilitated level of recognition efficiency. The masking effect of the noise mask did not vary significantly with the coarseness of quantization. The results supported neither a local feature processing account, nor a generalized spatial-frequency processing account, but were consistent with the microgenetic configuration-processing theory of face recognition. Also, the suitability of a spatial quantization technique for image configuration processing research has been demonstrated.