PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
 
Vision Res. Author manuscript; available in PMC 2010 November 1.
Published in final edited form as:
PMCID: PMC2783877
NIHMSID: NIHMS139921

Contrast polarity differences reduce crowding but do not benefit reading performance in peripheral vision

Abstract

Previous studies have shown that the spatial extent of crowding in peripheral vision is reduced when a target letter and its flanking letters have opposite contrast polarity. We have examined if this reduction in crowding leads to improved reading performance. We compared the spatial extent of crowding, visual-span profiles (plots of letter-recognition accuracy versus letter position), and reading speed at 10 deg inferior visual field, using white letters, black letters, or mixtures of white and black letters, presented on a mid-gray background. Consistent with previous studies, the spatial extent of crowding was reduced when the target and flanking letters had opposite contrast polarity. However, using mixed contrast polarity did not lead to improvements in visual-span profiles or reading speed.

Keywords: reading, crowding, contrast polarity, visual span, peripheral vision

INTRODUCTION

Our ability to see the fine details of a target is better when the target is presented alone than when it is closely surrounded by other objects (e.g. Bouma, 1970; Townsend, Taylor & Brown, 1971). This phenomenon, referred to as crowding, is ubiquitous in spatial vision and has been reported for a variety of spatial tasks including letter recognition (e.g. Bouma, 1970; Flom, Weymouth & Kahneman, 1963), Vernier discrimination (e.g. Levi, Klein & Aitsebaomo, 1985; Westheimer & Hauske, 1975), orientation discrimination (e.g. Andriessen & Bouma, 1976; Westheimer, Shimamura & McKee, 1976) and face recognition (Louie, Bressler & Whitney, 2007; Martelli, Majaj & Pelli, 2005). Many theories have been developed to explain the origin of crowding. Am ong these theories include the “physics” of the stimulus (Hess, Dakin & Kapoor, 2000; Liu & Arditi, 2000), receptive field size (Flom et al., 1963), awry feature integration at a stage beyond detection (e.g. Chung, Levi & Legge, 2001; Levi, Hariharan & Klein, 2002; Pelli, Palomares & Majaj, 2004) and an insufficient resolution at the attention level (e.g. He, Cavanagh & Intriligator, 1996; Intriligator & Cavanagh, 2001). For a review on the affected tasks and explanations of crowding, refer to Levi (2008). Irrespective of the origin of crowding, the consensus of opinion is that crowding causes deleterious effects on vision, as evidenced by the general observation that visual recognition improves when crowding is reduced. As such, there is a great deal of interest on developing methods to reduce the effect of crowding on important visual tasks, in the hope that visual performance can be improved. One pertinent finding from previous studies is that the magnitude of crowding decreases when a target and the flanking elements become more dissimilar in stimulus attributes. Such attributes include spatial frequency (Chung et al., 2001), shape, color, luminance and contrast polarity (Kooi, Toet, Tripathy & Levi, 1994).

Our interest in crowding stems from our interest in understanding why reading is slower in peripheral vision than in central vision. Even when print is enlarged appropriately in the periphery so that size is not a limiting factor (Chung, Mansfield & Legge, 1998; Latham & Whitaker, 1996), and when oculomotor demands for reading are minimized using the rapid serial visual presentation (RSVP) paradigm to present text, reading remains slower in peripheral than in central vision (Chung et al., 1998; Latham & Whitaker, 1996; Rubin & Turano, 1994). Given that letter identification is a classic example of how crowding reduces performance (e.g. Bouma, 1970; Chung et al., 2001; Flom et al., 1963), and that the magnitude (Jacobs, 1979; Loomis, 1978) and spatial extent (Jacobs, 1979; Latham & Whitaker, 1996; Toet & Levi, 1992) of crowding increase with distance away from fovea, crowding has long been suggested as a likely factor contributing to slow reading in peripheral vision.

If crowding is indeed a limiting factor on peripheral reading speed, then reading should be faster once crowding is reduced or minimized. The classic study of Bouma (1970) established that the magnitude of crowding decreases with increased separation between adjacent letters. Might increasing the spacing between letters help reading in peripheral vision? Unfortunately, not. Increasing letter spacing does not lead to improved reading performance (Chung, 2002; Yu, Cheung, Legge and Chung, 2007). Any benefits due to increased spacing are offset by deficits due to the letters in widely spaced text being pushed further into visual periphery where the resolution is poorer and the spatial extent of crowding is broader (Pelli et al., 2007).

In this study we explored another avenue that may help reading performance in peripheral vision while preserving the letter spacing — one that capitalizes on the observation that the spatial extent of crowding is reduced when a target and its flanking elements are made to be dissimilar (Kooi et al., 1994). Specifically, we studied one of these stimulus manipulations — the contrast polarity of letters — to test if reading speed improves for text with mixed contrast-polarity letters, compared with text with uniform contrast-polarity letters. Previous studies have shown that the use of opposite contrast polarity for a target letter and its flankers is effective in reducing the spatial extent of crowding (Kooi et al., 1994; Chakravarthi & Cavanagh, 2007, but see Hess, Dakin, Kapoor & Tewfik, 2000). In relation to reading, mixing contrast polarity has no impact on reading performance in central vision (Beckmann, Legge & Luebker, 1991) but to our knowledge, reading with mixed polarity text has not been assessed in peripheral vision. Given that the visual span has been suggested as the bottleneck on reading speed (Legge, Cheung, Yu, Chung, Lee & Owens, 2007) and that the visual span itself could be limited by crowding (Legge, 2007; Pelli et al., 2007), we also compared the visual spans for reading mixed vs. uniform contrast polarity text.

METHODS

To examine whether reading speed in peripheral vision can be improved when crowding is reduced using mixed-polarity text, we first validated the result that the spatial extent of crowding is smaller when a target and its flankers are rendered in opposite contrast polarity than when they are rendered in the same contrast polarity (Kooi et al, 1994). Then we compared reading performance for uniform- and mixed-contrast-polarity text. To link crowding, reading and visual span, we also compared the visual span using same and opposite contrast-polarity trigrams (strings of three random letters).

Stimuli and Apparatus

a. Spat ial extent of crowding

The stimulus used for determining the spatial extent of crowding consisted of a letter T surrounded by four flanking Ts. Each letter T subtended 0.5° × 0.5° with a 6 arc min line width as in the study of Kooi et al. (1994), see Figure 1. Previous studies showed that the magnitude and extent of crowding in peripheral vision are independent of stimulus size (Levi et al., 2002; Tripathy & Cavanagh, 2002), therefore here we only tested one letter size (but see control experiment in which we tested an observer using 3° letters). The letter Ts were either black (2 cd/ m2) or white (98 cd/ m2), presented against a mid-gray background (50 cd/ m2). We tested four combinations of contrast polarities of the target and its flanking Ts: white target with white flankers, black target with black flankers, white target with black flankers and black target with white flankers (see Figure 1, panels a–d). The target T was always presented at the center of the monitor with the four flanking Ts positioned equidistant from the target T. A small red dot serving as the fixation target, was presented 10 deg above the center of the target T. The orientation of each T (target or flankers) was chosen randomly from the four possible orientations: up, down, right or left. Stimuli were generated on a Macintosh 8600/ 300 computer using custom-written software, and were displayed on an Apple Display monitor (model number M4552) at a resolution of 1024 by 768 pixels and a frame rate of 75 Hz. Testing was conducted at 40 cm with each pixel subtending 2.54 arc min. Given that different readers read at different speeds and have different fixation durations during reading, to better link the extent of crowding with reading performance, we measured the spatial extent of crowding for three stimulus durations: 53, 147 and 1000 ms. In this way the data will reflect the extent of crowding for a range of fixation durations that might occur when reading using peripheral vision.

Figure 1
Sample stimuli for the experiment. The top row shows the stimulus configuration used for measuring the spatial extent of crowding. In each panel, a four-orientation target T was surrounded by four other flanking Ts. Panels a and b show the Ts rendered ...

In each block of 140 trials, we used the Method of Constant Stimuli to assess the response accuracy of identifying the orientation of the target T at six different target-flanker separations, as well as the unflanked condition. The six separations varied for different conditions but all spanned a range of separations such th at observer‘s performance approximately ranged from chance (25% correct) to 100% correct. Each separation was tested 20 times. On each trial, a letter T, with or without the four flanking Ts, was presented for one of the three stimulus durations (different durations were tested in different blocks of trials). Observers‘ task was to indicate, using a keyboard, the orientation of the target T. Feedback was not provided. For each condition, we fit a cumulative-Gaussian function to relate the response accuracy with target-flanker separation (see Figure 2). From each fitted function, we derived the spatial extent of crowding, defined as the letter separation that yielded 62.5% correct response (50% for a four-alternative forced-choice task).

Figure 2
Samples of psychometric functions from observer S3 relating proportion correct of identifying the orientation of T stimuli with target-flanker (center-to-center) separation (in deg). The rightmost data points plotted in each panel, offset slightly to ...

Control data were collected using 3° T stimuli (matching the letter size used in the RSVP task and similar to the letter size used in the visual span measurement) and with only horizontal flankers (left and right), to closer mimic the stimulus configuration of the reading and visual-span tasks. See Discussion for details.

b. Reading speed

Oral reading speed for single sentences was measured using the rapid serial visual presentation (RSVP) paradigm, i.e. words were presented one at a time in rapid succession, each for a fixed exposure duration (e.g. Chung et al., 1998; Rubin & Turano, 1992; 1994). In addition to testing at 10 deg in the inferior visual field, we also measured reading speeds at the fovea and 5 deg in the inferior visual field to see if the contrast polarity effect on reading changes with retinal eccentricity. Stimuli were generated using an SGI O2 workstation (Silicon Graphics Inc.), and presented on a Sony color graphics display monitor (model# GDM-17E21, refresh rate = 75 Hz) controlled by custom-written software. The psychophysical procedures and sentence set used for measuring reading speed were identical to those used by Chung et al. (1998) and Chung (2002). In brief, on each trial, a single sentence was chosen randomly, without replacement, from a pool of 2630 sentences selected from classic literature. Each sentence contained between 8 and 14 words (mean = 11 ± 1.7 words). All the words were among the 5000 most frequent words in written English, according to word-frequency tables derived from the British National Corpus (Kilgarriff, 1997). We used the Method of Constant Stimuli to present words at six exposure durations that spanned a range of approximately 1 log unit (e.g. 80, 133, 213, 333, 533, 800 ms per word). Viewing distances were 115 cm for foveal testing and 40 cm for testing at 5 and 10 deg eccentricity. Print sizes were 0.34°, 1.5° and 3.0° for measurements at the fovea, 5 and 10 deg eccentricity, respectively. These sizes were twice the average critical print sizes (the smallest print size that allows observers to read at their maximum reading speed) at the respective eccentricity as reported in Chung et al. (1998). Text was displayed using the Times-Roman font. Observers were asked to read each sentence as quickly and as accurately as possible while fixating on a fixation line, if present. The number of words read correctly was recorded for each trial. The experimenter visually monitored the observer‘s eyes, to verify that the observers maintained fixation on the line. Horizontal eye movements along the fixation line were allowed, although observers usually preferred fixating at a certain location on the fixation line instead of scanning along the horizontal fixation line. A trial was discarded and repeated with a different sentence when vertical eye movements were detected. Averaged across observers, approximately 12% of trials were discarded and repeated. This trial rejection rate was similar to that reported by Chung et al. (1998) where an eye-tracker was used to monitor observers‘ fixation.

Reading speeds were measured for four text conditions. In all conditions, letters were presented on a mid-gray background (55 cd/ m2). The four conditions consisted of (1) all white (108 cd/ m2) letters, (2) all black (2 cd/ m2) letters, (3) interleaved white and black letters and (4) randomly mixed white and black letters (see Figure 1, panels e –h). Data obtained for each text condition were tallied for each observer. We fit a cumulative-Gaussian function to relate the proportion of words read correctly as a function of word exposure duration. Each curve was based on 36 sentences (six sentences tested for each of the six durations). Reading speed was calculated from these curves as the RSVP presentation rate (in words per minute) that yielded 80% of words read correctly, as in previous studies (e.g. Chung et al., 1998; Chung, 2002; 2007; Chung, Legge & Cheung, 2004, Yu et al., 2007).

c. Visual span profile measurement

Visual span profiles, defined as plots of letter recognition accuracy as a function of letter position left and right of the midline, were measured using a letter recognition task, as described in Legge, Mansfield and Chung (2001). In brief, on each trial, a trigram (a sequence of three lowercase letters randomly chosen from the 26 letters) was presented for 147 ms along a horizontal line that was 10 deg below the fixation target. We tested trigrams at 11 positions (indexed by the position of the middle letter) from five letter spaces to the left of fixation to five letter spaces to the right of fixation (20.3 deg on either side of fixation). A trigram centered on the midline assumes a position 0. Each trigram position was tested 20 times. Letter size was 3.5°, identical to the letter size used in Legge et al. (2001) for testing at 10 deg in the inferior visual field. Letters were rendered in Courier,1 as in previous studies (Chung et al., 2004; Legge et al., 2001). Stimuli were generated using an SGI O2 workstation (Silicon Graphics Inc.), and presented on a Sony color graphics display monitor (model# GDM-17E21, refresh rate =75 Hz) controlled by custom-written software. Viewing distance was 40 cm. Observers reported the identity of the three letters, from left to right. Feedback was not provided to the observers. A letter was scored as correct if and only if its order within the trigram was also correct. Visual-span profiles were constructed by calculating the proportion of letters that were correctly identified at each letter position. These calculations combined data across cases in which the letter position contained the middle, left, or right component of the trigram.

Visual span profiles were measured for three conditions that differed in the color of the letters presented on a gray background: (1) all letters were white, (2) all letters were black and (3) letters alternated in black and white (see Figure 1, panels i–l).

Observers

Three native English speakers with normal vision aged between 22 and 28 participated in all three parts of this study. All had (corrected) visual acuity of 20/ 16 or better in each eye and were either emmetropic or wore corrective lenses to correct for refractive errors. Testing was binocular. None of the observers was aware of the purpose of the experiment. Written informed consent was obtained from each observer after the procedures of the experiment were explained and before the commencement of data collection. For each part of the study the observers were given extensive practice trials (typically for an hour) in order to become familiar with the experimental task and to obtain asymptotic performance.

Data for the control conditions were collected from two observers (aged 20 and 22) with normal vision who did not participate in the main experiment and were unaware of the purpose of the testing.

RESULTS

Sample psychometric functions relating the accuracy of identifying the orientation of the T stimuli as a function of target-flanker (center-to-center) separation are presented in Figure 2, for the three stimulus durations and the four combinations of target and flanker contrast polarities. Performance for identifying the unflanked T stimuli is also included. From these psychometric functions (and others that are not shown here), we derived the spatial extent of crowding as the letter separation that yielded 62.5% correct on these psychometric functions. The spatial extent of crowding, plotted as a function of stimulus duration, is compared for the four conditions of different combinations of target and flanker contrast polarities in Figure 3. Each panel presents data for one observer. Despite individual differences in the absolute extent of crowding, our data show a few characteristics of crowding that are consistent across observers. First, in good agreement with a previous finding, the spatial extent of crowding shrinks with increased stimulus duration (Tripathy & Cavanagh, 2002). Second, the spatial extent of crowding is larger when the target and flankers share the same contrast polarity, and smaller when the contrast polarity of the target and flankers are different. This effect occurs for a range of stimulus durations. Third, there is practically no difference (repeated measures ANOVA: F(df=1,2) = 2.61, p = 0.25) in the spatial extent of crowding whether the target and flankers were all black or all white, for the same-polarity conditions. Similarly, there is also practically no difference (repeated measures ANOVA: F(df=1,2) = 1.00, p = 0.42) in the spatial extent of crowding for a white target with black flankers, or a black target with white flankers, for the opposite-polarity conditions. Figure 4 shows the spatial extent averaged across observers and the two respective conditions for the same- and opposite contrast polarity conditions. In general, the spatial extent of crowding is larger for the same- than the opposite contrast-polarity conditions by factors of 1.87, 2.20 and 2.07, for stimulus durations of 53, 147 and 1000 ms, respectively.

Figure 3
The spatial extent of crowding (deg) is plotted as a function of stimulus duration (ms) for the four contrast polarity conditions and for the three observers. Measurements were obtained at 10 deg in the inferior visual field. The figure legend shows the ...
Figure 4
The spatial extent of crowding (deg), averaged across the three observers and two conditions, is plotted as a function of stimulus duration (ms) for the same- and opposite-contrast polarity conditions. Error bars represent ± 1 S.E.M., taking into ...

Figure 5 compares how RSVP reading speed changes with eccentricity for the four conditions of text (letter) contrast polarity. Consistent with previous studies (Chung et al., 1998; Latham & Whitaker, 1996), reading speed is highest at the fovea and drops with increased eccentricity. The averaged rate of change of RSVP reading speed with eccentricity is highly comparable to what we reported earlier (Chung et al., 1998). The important point, however, is that reading speed is virtually identical for the four text contrast polarity conditions (repeated-measures ANOVA: F(df = 3,6) = 0.39, p = 0.77). In other words, text with mixed contrast polarity (interleaved or the random conditions) does not have a benefit in reading speed over text with uniform contrast polarity (all white or all black letters). This effect is observed at the fovea, as well as in the periphery.

Figure 5
RSVP reading speed (wpm) is plotted as a function of eccentricity (deg) for the four text polarity conditions (see legend), and for the three observers. Error bars represent ± 1 S.E.M., estimated from the p sychometric function fit to each data ...

In an attempt to reconcile the findings that mixed contrast polarity stimuli yielded a reduction in the spatial extent of crowding without benefiting reading speed, we measured visual span profiles for trigrams with different contrast polarity conditions. Figure 6 shows the visual span profiles — proportion correct of letter identification as a function of letter position, for the three contrast polarity conditions. In general, the visual span profiles show the expected inverted-U shape function, with highest performance accuracy at letter position 0, corresponding to midline, and a reduction in accuracy with increased letter position away from fixation. However, just like for reading speed, the visual span profile does not depend on the contrast polarity of the individual letters of trigrams. To quantify the comparison, we fit a split-Gaussian curve to describe each set of visual span data, as in our previous studies (Legge et al., 2001; Chung et al., 2004). The split Gaussian can be characterized by three parameters: the overall amplitude and the standard deviations of the left and right side of the curve. Another way to characterize a visual span profile is to quantify the size of the visual span by summing across the information transmitted in each slot (akin to computing the area under the visual span profile). For details of the transformation from letter-recognition accuracy to bits of information, see Legge et al (2001). A repeated-measures ANOVA shows that none of the four parameters of the visual span profile depends on the contrast polarity of the stimuli (Greenhouse-Geisser corrected p-values: 0.86 for the amplitude; 0.80 for the left standard deviation; 0.38 for the right standard deviation and 0.55 for the bits of information transmitted).

Figure 6
Proportion correct of letter identification is plotted as a function of letter position for the three trigram polarity conditions, and for the three observers. Error bars represent ± 1 S.E.M., obtained using bootstrapping with 300 simulations. ...

DISCUSSION

The goal of this study was to test whether mixed contrast polarity in text, which has been shown to reduce the spatial extent of crowding, leads to improved reading speed in peripheral vision. We reasoned that if crowding limits peripheral reading, then by reducing the spatial extent of crowding, reading speed in peripheral vision should improve. Contrary to our prediction, the use of mixed-polarity text does not improve reading speed in normal peripheral vision. Our failure to find an improvement in reading speed using mixed-polarity text is reflected in our measurements of the visual span which also demonstrate no difference for same vs. mixed-polarity trigram stimuli. Legge et al. (2007) have shown that changes in reading speed caused by experimental manipulation of print size, letter contrast, retinal eccentricity, and perceptual learning, are positively correlated with changes in the size of the visual-span under the same conditions. Our finding that reading speed and visual-span size are not affected by manipulating contrast polarity is consistent with Legge et al.‘s proposition that there is a direct link between reading speed and the size of the visual span.

This leaves the question of why mixing contrast polarity reduces the spatial extent of crowding using the oriented-T task, but does not improve the visual span or reading performance. In the following sections we will consider factors that could explain this puzzling result.

a) Stimulus configuration

It is possible that the differences in the impact of contrast polarity among these tasks could be explained by the differences in the configuration between the oriented T stimuli and the trigram and reading stimuli. For example, the oriented Ts were small, 0.5°, and close to the size threshold whereas the trigram and reading stimuli were considerably larger, 3–3.5° (large enough to support maximum reading speed at 10° in the visual periphery). Also, the stimuli in the oriented-T task were flanked horizontally and vertically (left-right, and top-bottom) whereas the stimuli in the visual span and reading tasks were affected by crowding (from the other letters in words, or from adjacent letters in trigrams) only in the horizontal direction. To test if the effect of contrast polarity differences on the spatial extent of crowding depended on the stimulus size and the number and location of flankers, we collected additional data from two observers (S4 & S5) using 3° T-stimuli (matching the letter size used in the RSVP task, and close to the letter size for visual-span measurement) and with only two flankers — one on the right and one on the left of the target T. The target T was 10° below and 12° to the right of fixation. This location corresponds to 3–4 letters to the right of the midline for the visual span measurement, and to the third or fourth letter of a word in the reading speed measurement.2 Presentation duration was 300 ms, close to the word-presentation duration for threshold reading performance of approximately 200 wpm at 10° eccentricity (see Figure 5). Even with these large stimuli and with only two flankers, we still found a substantial effect of polarity differences on the spatial extent of crowding (Figure 7). Compared with the psychometric functions shown in Figure 2, it is clear that the magnitude of crowding was substantially reduced for the larger 3° T stimuli (Figure 7) than the smaller 0.5° stimuli. The important point, however, is whether the contrast polarity advantage persists for larger stimuli. For the range of separations tested, same-polarity flankers showed a marked detrimental effect on the performance-accuracy of identifying the orientation of the target Ts, whereas mixed-polarity flankers practically had no effect on the performance-accuracy. These data clearly indicate that mixed-polarity flankers are still effective in reducing crowding for large stimuli.

Figure 7
Proportion correct of identifying the orientation of the T stimuli is plotted as a function of the target-flanker separation (in deg), for 3° T stimuli. Two conditions were compared: white target Ts flanked by white Ts (“WWW”, ...

b) Factors affecting the visual span

According to Legge (2007), the size of the visual span is determined by (1) the decrease in acuity outward from the midline, (2) the reduction in accuracy of position signals in peripheral vision, and (3) crowding between adjacent letters. With respect to acuity, Westheimer, Chu, Huang, Tran & Dister (2003) showed that for people in the same age-range as the three observers in this study, acuities for black-on-white or white-on-black optotypes are very similar. Their measurements were obtained at the fovea. To our knowledge, contrast polarity effects on acuity in the periphery have not been investigated previously but in general, negative polarity (e.g. white-on-black) stimuli seem to benefit people who suffer from significant amount of intraocular scattering of light, as in patients with cataracts (Legge, Rubin, Pelli & Schleske, 1985;Westheimer, 2003; Westheimer & Liang, 1995). Since there is no a priori reason to believe that intraocular scattering increases with off-axis viewing, we assume that changes in acuity with distance from fixation should have very minimal dependence on the contrast polarity of the letters.

As for position signals, it is well documented that when two objects are separated by a gap larger than a few arc min, the localization of their positions is mediated by local signs, akin to position tags (Waugh & Levi, 1993). For our trigrams, adjacent letters are separated from one another by more than a few arc min. Therefore, it is likely that the position signals of letters are mediated by local signs (Chung & Legge, 2009). A characteristic of the local sign mechanism is that the accuracy of position judgment is independent of the contrast polarity of the objects — the accuracy is similar whether the two objects share the same or opposite contrast polarity (e.g. Levi & Waugh, 1996; Levi & Westheimer, 1987; O‘Shea & Mitchell, 1990).

The only factor among the three proposed by Legge (2007) that seems to be susceptible to the contrast polarity effect is crowding between adjacent letters. We have demonstrated that the spatial extent of crowding is smaller for mixed-polarity T-stimuli. At first glance it is surprising that this reduction in the spatial extent of crowding does not lead to improved visual span measurements and reading performance. However, the crowding effect can be quantified by at least two parameters — its spatial extent and its magnitude. The spatial extent refers to the lateral spread of the interaction effect, and is usually quantified according to some criteria, e.g. the separation between a target and its flankers that yields a certain percent-correct of identification accuracy of the target (we used 62.5% in our study). The magnitude usually refers to the maximum decrement in performance due to the presence of flankers. Even though the spatial extent of crowding is reduced when the target and flankers have mixed contrast polarities, perhaps the magnitude of crowding is similar for uniform- and mixed-polarity stimuli. This is not the case for our study, however. Using the psychometric functions relating the proportion correct of identifying the orientation of the T-stimuli with target-flanker separation (examples shown in Figure 2), we determined the decrement in performance (1 – proportion correct performance) at a target-flanker separation of 4 deg, equivalent to the separation between adjacent letters for the trigram stimuli. Table 1 lists the spatial extent and the magnitude of crowding, averaged across the three observers, for the three stimulus durations. The relative sign of contrast polarity between the target and the flanking Ts affects not only the spatial extent, but also the magnitude of crowding. When the target and the flanking Ts have opposite contrast polarity, both the spatial extent and the magnitude of crowding become smaller.

Table 1
Averaged values (± 1 SEM) of the spatial extent (deg) and magnitude of crowding when the target and flanking T stimuli had the same or opposite contrast polarity. The magnitude of crowding was obtained at a spatial extent of 4 deg.

c) Task differences

In the oriented-T task, observers could use a strategy of ignoring the flanking Ts, a tactic that may be made easier when the flankers and target have opposite contrast polarity. However, such a strategy could not be employed in the trigram and RSVP tasks because the observers were required to identify all three letters in the trigrams, and enough letters to identify each word when reading. To investigate if such a strategy could contribute to the difference in the contrast polarity advantage observed for the different tasks, we tested observers S4 and S5 using trigram stimuli (3.5° letter size, as in the main study) presented at 10° in the inferior visual field and with the middle letter positioned 3 letter slots right of the vertical midline. Stimulus presentation duration was 300 ms, to match the control data collected for the oriented-T task using 3° letters and threshold reading speed using RSVP (approximately 200 wpm). The task, however, differed from the visual span measurement in the main study in that we asked the observers to only report the identity of the middle letter. We compared two contrast polarity conditions: white target with white flankers and white target with black flankers. One hundred trials were tested for each condition for each observer. Our prediction was that if the observer s only need to identify the middle letters of trigrams, then the performance accuracy should be higher for mixed-polarity trigrams than for same-polarity trigrams, just like what we observed for the oriented-T task. Consistent with our prediction, the proportion correct for identifying the middle letters of mixed-polarity trigrams averaged 0.82 while that for same-polarity trigrams averaged 0.61. In contrast, when we reanalyzed the visual-span trials from the main experiment to consider only the performance accuracy when the same letter position (3 letter slots right of the vertical midline) was occupied by the middle letters of trigrams (i.e. trials for which the letter position was occupied by the left or right component of the trigrams were excluded for analysis), there was only a very small difference in performance accuracy between the mixed and same-polarity trigrams (0.50 vs. 0.44). Contrast polarity differences help greatly when the task is to identify a single flanked letter, but do not help so much when the task is to identify multiple letters. This result implies that the difference in the contrast polarity advantage observed for the oriented-T, RSVP reading and the visual span tasks may be due, at least in part, to whether observers are allowed to ignore the flankers.

Conclusions

In summary, we have demonstrated that the spatial extent of crowding is reduced when the target and flanking letters had opposite contrast polarity. However, using mixed contrast polarity does not lead to improvements in visual-span profiles or reading speed. The difference in how contrast polarity affects these tasks may be in part due to the different task demands of the oriented-T task versus the identification of multiple letters in trigrams and words.

From a practical point of view, our quest for a simple text manipulation to improve peripheral reading speed remains.

Acknowledgments

This study was supported by research grants EY012810 and EY002934 from NIH. We thank Dr. Gordon Legge for his insightful comments and suggestions. Portions of this study were presented at the Vision99 conference in New York, 1999.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

1The different fonts used in each part of the study match the fonts that are typically used for these types of task. The spatial extent of crowding task requires an optotype that has the same width when rotated by 90°. The visual span measurements use a fixed-width font, such as Courier, in order to simplify the measurement of letter-recognition accuracy for each letter position across the span. Most everyday reading tasks use a proportionally spaced fonts, such as Times. Reading performance only differs slightly between fonts provided the print size is larger than the critical print size (Mansfield, Legge & Bane, 1996) and thus it is unlikely that font differences will have a major impact on our findings.

2In the main study, the spatial extent of crowding was 3.6° at the original location for 147 msec durations when we used four flankers. We anticipated that the extent would be smaller for 300 ms durations using only two flankers, and so, to avoid a ceiling effect with the larger Ts, we tested at a location 12° to the right of the midline where the zone of crowding is larger.

REFERENCES

  • Andriessen JJ, Bouma H. Eccentric vision: Adverse interactions between line segments. Vision Research. 1976;16:71–78. [PubMed]
  • Beckmann PJ, Legge GE, Luebker A. Reading: Letters, words and their spatial frequency content. Society for Information Display 1991 Digest of Technical Papers. 1991:106–108.
  • Bouma H. Interaction effects in parafoveal letter recognition. Nature. 1970;226:177–178. [PubMed]
  • Chakravarthi R, Cavanagh P. Temporal properties of the polarity advantage effect in crowding. Journal of Vision. 2007;7(2)(11):1–13. [PubMed]
  • Chung STL. The effect of letter spacing on reading speed in central and peripheral vision. Investigative Ophthalmology and Visual Science. 2002;43:1270–1276. [PubMed]
  • Chung STL. Learning to identify crowded letters: Does it improve reading speed? Vision Research. 2007;47:3150–3159. [PMC free article] [PubMed]
  • Chung STL, Legge GE. Precision of position signals for letters. Vision Research. 2009;49:1948–1960. [PMC free article] [PubMed]
  • Chung STL, Legge GE, Cheung S-H. Letter recognition and reading speed in peripheral vision benefit from perceptual learning. Vision Research. 2004;44:695–709. [PMC free article] [PubMed]
  • Chung STL, Levi DM, Legge GE. Spatial-frequency and contrast properties of crowding. Vision Research. 2001;41:1833–1850. [PubMed]
  • Chung STL, Mansfield JS, Legge GE. Psychophysics of reading. XVIII. The effect of print size on reading speed in normal peripheral vision. Vision Research. 1998;38:2949–2962. [PubMed]
  • Flom MC, Weymouth FW, Kahneman D. Visual resolution and contour interaction. Journal of the Optical Society of America. 1963;53:1026–1032. [PubMed]
  • He S, Cavanagh P, Intriligator J. Attentional resolution and the locus of visual awareness. Nature. 1996;383:334–337. [PubMed]
  • Hess RF, Dakin SC, Kapoor N. The foveal ‘crowding’ effect: Physics or physiology? Vision Research. 2000;40:365–370. [PubMed]
  • Hess RF, Dakin SC, Kapoor N, Tewfik M. Contour interaction in fovea and periphery. Journal of the Optical Society of America A. 2000;17:1516–1524. [PubMed]
  • Intriligator J, Cavanagh P. The spatial resolution of visual attention. Cognitive Psychology. 2001;43:171–216. [PubMed]
  • Jacobs RJ. Visual resolution and contour interaction in the fovea and periphery. Vision Research. 1979;19:1187–1195. [PubMed]
  • Kilgarriff A. Putting frequencies in the dictionary. International Journal of Lexicography. 1997;10:135–155.
  • Kooi FL, Toet A, Tripathy SP, Levi DM. The effect of similarity and duration on spatial interaction in peripheral vision. Spatial Vision. 1994;8:255–279. [PubMed]
  • Latham K, Whitaker D. A comparison of word recognition and reading performance in foveal and peripheral vision. Vision Research. 1996;36:2665–2674. [PubMed]
  • Legge . Psychophysics of reading in normal and low vision. Mahwah, NJ: Lawrene Erlbaum Associates, Inc.; 2007.
  • Legge GE, Cheung S-H, Yu D, Chung STL, Lee H-W, Owens DP. The case for the visual span as a sensory bottleneck in reading. Journal of Vision. 2007;7(2)(9):1–15. http://journalofvision.org/7/2/9/ [PMC free article] [PubMed]
  • Legge GE, Mansfield JS, Chung STL. Psychophysics of reading. XX. Linking letter recognition to reading speed in central and peripheral vision. Vision Research. 2001;41:725–743. [PubMed]
  • Legge GE, Rubin GS, Pelli DG, Schleske MM. Psychophysics of reading: II. Low vision. Vision Research. 1985;25:253–265. [PubMed]
  • Levi DM. Crowding – An essential bottleneck for object recognition: A mini-review. Vision Research. 2008;48:635–654. [PMC free article] [PubMed]
  • Levi DM, Hariharan S, Klein SA. Suppressive and facilitatory spatial interactions in peripheral vision: Peripheral crowding is neither size invariant nor simple contrast masking. Journal of Vision. 2002;2:167–177. [PubMed]
  • Levi DM, Klein SA, Aitsebaomo AP. Vernier acuity, crowding and cortical magnification. Vision Research. 1985;25:963–977. [PubMed]
  • Levi DM, Waugh SJ. Position acuity with opposite-contrast polarity features: Evidence for a nonlinear collector mechanism for position acuity? Vision Research. 1996;36:573–588. [PubMed]
  • Levi DM, Westheimer G. Spatial interval discrimination in the human fovea: What delimits the interval? Journal of the Optical Society of America, A. 1987;4:1304–1313. [PubMed]
  • Liu L, Arditi A. Apparent string shortening concomitant with letter crowding. Vision Research. 2000;40:1059–1067. [PubMed]
  • Loomis JM. Lateral masking in foveal and eccentric vision. Vision Research. 1978;18:335–338. [PubMed]
  • Louie EG, Bressler DW, Whitney D. Holistic crowding: Select ive interference between configural representations of faces in crowded scenes. Journal of Vision. 2007;7(2)(24):1–11. [PMC free article] [PubMed]
  • Mansfield JS, Legge GE, Bane MC. Psychophysics of reading. XV. Font effects in normal and low vision. Investigative Ophthalmology and Visual Science. 1996;37:1492–1401. [PubMed]
  • Martelli M, Majaj NJ, Pelli DG. Are faces processed like words? A diagnostic test for recognition by parts. Journal of Vision. 2005;5:58–70. [PubMed]
  • O‘Shea RP, Mitchell DE. Vernier acuity with opposite-contrast stimuli. Perception. 1990;19:207–221. [PubMed]
  • Pelli DG, Palomares M, Majaj NJ. Crowding is unlike ordinary masking: Distinguishing feature integration from detection. Journal of Vision. 2004;4:1136–1139. [PubMed]
  • Pelli DG, Tillman KA, Freeman J, Su M, Berger TD, Majaj NJ. Crowding and eccentricity determine reading rate. Journal of Vision 7. 2007;7(2)(20):1–36. [PubMed]
  • Rubin GS, Turano K. Reading without saccadic eye movements. Vision Research. 1992;32:895–902. [PubMed]
  • Rubin GS, Turano K. Low vision reading with sequential word presentation. Vision Research. 1994;34:1723–1733. [PubMed]
  • Toet A, Levi DM. The two-dimensional shape of spatial interaction zones in the parafovea. Vision Research. 1992;32:1349–1357. [PubMed]
  • Townsend JT, Taylor SG, Brown DR. Lateral masking for letters with unlimited viewing time. Perception and Psychophysics. 1971;10:375–378.
  • Tripathy SP, Cavanagh P. The extent of crowding in peripheral vision does not scale with target size. Vision Research. 2002;42:2357–2369. [PubMed]
  • Waugh SJ, Levi DM. Visibility and vernier acuity for separated targets. Vision Research. 1993;33:1619–1638. [PubMed]
  • Westheimer G. Visual acuity with reversed -contrast charts: I. Theoretical and psychophysical investigations. Optometry and Vision Science. 2003;80:745–748. [PubMed]
  • Westheimer G, Hauske G. Temporal and spatial interference with vernier acuity. Vision Research. 1975;15:1137–1141. [PubMed]
  • Westheimer G, Liang J. Influence of ocular light scatter on the eye’s optical performance. Journal of the Optical Society of America, A. 1995;12:1417–1424. [PubMed]
  • Westheimer G, Chu P, Huang W, Tran T, Dister R. Visual acuity with reversed-contrast charts: II. Clinical investigation. Optometry and Vision Science. 2003;80:749–752. [PubMed]
  • Westheimer G, Shimamura K, McKee SP. Interference with line-orientation sensitivity. Journal of the Optical Society of America. 1976;66:332–338. [PubMed]
  • Yu D, Cheung S-H, Legge GE, Chung STL. Effect of letter spacing on visual span and reading speed. Journal of Vision 7. 2007;2(2):1–10. [PMC free article] [PubMed]