Color and saturation effects on perception: the Hermann Grid.
Much of the previous research that has been conducted on the Hermann Grid has focused on developing an explanation for the perception of the ID that occurs when one looks at the grid (Schiller & Carvey, 2005). Oehler and Spillman (1981) attribute the primary explanation of the Hermann Grid to Baumgartner in 1960. Baumgartner claimed perception of the ID was the result of lateral inhibition in the receptive fields of ganglion cells. A visual receptive field is the portion of the visual field that must be stimulated for a given neuron to fire. These fields are typically circular, composed of an inner circle and an outer ring. The response to stimulation of the inner circle is the opposite of the response to stimulation of the outer ring, so that an "on-center/off-surround" receptive field sends an excitatory response with stimulation of the inner circle and an inhibitory response to stimulation of the outer ring. In an "off-center/on-surround" an inhibitory response would be sent to the ganglion cell layer in response to stimulation of receptors in the center of the field, and an excitatory response would be sent when the surrounding ring is stimulated. Lateral inhibition is a competitive interaction between the center and the surround part of the receptive field. Perception can be complicated by simultaneous stimulation of both center and surround, such as what occurs when you look at the intersection of the white bands of the Hermann Grid. At the intersection of the white bands the inhibitory surround of an on-center/off-surround receptive field receives more inhibition than does the excitatory center of the field. This results in the perception of a "less-than white" (grey) dot at the intersection of the bars. Away from the intersection of the white bars, the inhibitory surround receives less stimulation than does the excitatory center, resulting in the perception of a white bar (see Figure 2). Goldstein (2010) provides an excellent description both of the organization of retinal receptive fields and of lateral inhibition (pages 45-47).
[FIGURE 1 OMITTED]
Several recent studies have suggested that Baumgartner's explanation of the illusion as dependent on lateral inhibition in the retina is not supported by the data (Cox, Ares-Gomez, Pacey, Gilchrist & Mahalingam, 2007; Schiller & Carvey, 2005; Vergeer & van Lier, 2010). Altering the straightness of the lines of the grid (Geier, Bernath, Hudak & Sera., 2004), and the orientation of the grid (De Lafuente & Ruiz, 2004) do not change the amount of lateral inhibition in the signal and so ought not to affect the perception of the ID, yet both manipulations of the grid eliminate the illusion. Increasing the size of the grid (Schiller & Carvey, 2005) and adding diagonal white lines cross the squares of the grid (Lingelbach, Block, Hatzky & Reisinger (1985) cited in Schiller & Carvey, 2005) increase the amount of lateral inhibition and so ought to increase the apparent darkness of the ID, yet neither manipulation does so, and adding the diagonal line actually eliminates the illusion for most viewers.
[FIGURE 2 OMITTED]
Researchers have also examined the role of color with regard to the perception of the Hermann Grid. Oehler and Spillman (1981) investigated color in the Hermann Grid by changing the color of the bars and background. The researchers found that the strongest illusion occurs when the bars and the background were the same color. Oehler and Spillman (1981) also claimed that color is interpreted using the red and green cones, with little input from the blue cones. This implies that colors that involve blue cones, such as blue, green, and yellow, will be harder to perceive on the Hermann Grid. Levine, Spillman, and Wolf (1980) also examined color with regards to the Hermann Grid. These researchers changed the color of the bars and background and found that participants reported seeing a dot more frequently with the red, blue, and purple intersections than with the green and yellow bars.
The purpose of the present study is to determine if changing the color and saturation of the squares of the Hermann Grid will affect the perception of the ID. We predict that the color of the ID will correspond with the color of the Hermann Grid and that low saturation levels will make perceiving a dot more difficult than high levels of saturation. In the few studies that have manipulated the color of the background of the grid, researchers have found evidence that long wavelengths make perception of the ID more difficult. Accordingly, we hypothesized that most of the subjects will be unable to perceive the ID on a yellow or red Hermann Grid regardless of saturation level. Finally, because the illusion is eliminated when viewers look directly at the ID, we predict participants who perceive the ID will focus their gaze at part of the stimulus away from the intersection of the bars.
Fifty-one participants completed the study. All participants were female college-age (18 to 21 years of age) students. Participants were asked if they had ever been tested for color blindness and/or if they knew if they were colorblind. No participants were colorblind. Twenty-five participants had their vision corrected with glasses or contacts while completing the study.
The Hermann Grids were created using Microsoft's Paint program. The grids were six different colors (black, red, yellow, green, blue, and purple) with three different saturation levels for each of the colors (80,160,240 bits per pixel) for a total of 18 grids. The colored grids all had the same level of luminance (120 bits per pixel). Each square grid measured 6.35 by 6.35 cm.
A campus-wide e-mail was used to recruit participants. Participants signed-up for a time to come into the lab to complete the study. An explanation of what the Hermann Grid is, what the traditional layout of the grid looks like, along with a clarification about what dots the questions were referring to was provided to all participants before beginning. All participants viewed the 18 grids (Figure 3) individually on a computer screen. We used ClearView 2.7.0 eye-tracker software on an IBM Lenovo ThinkCentre desktop with a dual monitor setup to track eye-movements and gaze location while participants viewed the grids. Participants viewed the stimuli on a Radeon X300 series eye-tracker monitor. The monitor had a resolution of 1280 by 1024 pixels and was set to the highest color quality available, 32 bit. Participants were calibrated to the eye-tracker before viewing the stimuli in order to optimize the computer's ability to track her eyes. They were given a maximum of 45 seconds to view each grid and answer three questions:
[FIGURE 3 OMITTED]
Do you see a dot? What color is the dot? What is the level of clarity of the dot? Participants were allowed to choose very faint, moderately faint, moderately clear, or very clear to answer the last question. The questions were read out loud to the participants, so that the participant could continue looking at the grid while answering. The researcher began asking the series of questions after 30 seconds if the participant had not started to reply before the 30 second mark. Answers were recorded and later coded.
The first hypothesis stated that the color of the ID will correspond with the color of the Hermann Grid. Figure 4 shows the responses to the question about the color of the ID. Most participants reported that the color of the ID did not correspond to the color of the grid, except for the black grids, and the highest saturated red and yellow grids. The most frequent ID color reported, was grey.
[FIGURE 4 OMITTED]
The second hypothesis stated that low saturation levels will make perceiving a dot more difficult than high levels of saturation. A two-way Chi-Square Test of Independence found that perception of the ID was dependent on saturation level, [chi square] = 33.585, p < .001. After the data were split by color and the same two-way test of independence was performed for each of the six colors, we found that only the yellow stimuli showed a significant dependency between perception of the ID and saturation, [chi square] = 48.402, p < .001. As saturation increased, fewer participants reported seeing the ID, for yellow stimuli alone. Figure 5 shows the responses to the question "Do you see a dot?" across saturation level and color of the grid.
The third hypothesis stated that most of the subjects will be unable to perceive the ID on a yellow Hermann Grid regardless of saturation level. To test this hypothesis, we first split the data by saturation level and used a two-way Chi-square test of independence to compare whether perception of the ID was associated with the color of the stimulus. We found that the color of the grid and seeing the ID were dependent at the middle saturation level, [chi square](5) = 56.581, p < .001, and at the highest saturation level, [chi square](5) = 129.123, p < .001. We then eliminated the yellow grid data from the data set and repeated the analysis. With yellow removed from the data, the results were no longer statistically significant, suggesting that perception of the ID was eliminated on the yellow grid alone.
[FIGURE 5 OMITTED]
A two-way Chi-Square Test of Independence was used to test the last hypothesis about gaze location. At all three saturation levels, participants were more likely to look at the non-white areas (the colored squares and/or the edge of the colored squares) of the Hermann Grid than at the white areas (the white bars between the squares). Saturation level and gaze location were found to be independent, [chi square](2) = .617, p > .05. We also compared the effect of color on gaze location, and found that gaze location was dependent on color, [chi square](5) = 11.43, p < .05, regardless of saturation level (see Figure 6). For all colors presented, except for yellow, gaze was more frequently located in a non-white area of the stimulus. For the yellow grids, gaze was split almost evenly between the white and non-white areas.
The results of the study supported our hypothesis that perception of the ID occurred when gaze was directed away from the intersection of the white background bars. The only exception to this result occurred with yellow grids. When the stimulus was yellow, participants tended to look at both the white intersections and the colored grids equally often. We found that high saturation levels resulted in poor perception of the ID. Perception of the ID was significantly better at low saturation levels than at higher levels. Our hypotheses that most participants would see an ID that was the same color as the grid was not supported. In fact, most participants reported that they saw an ID that was a different color than the grid (usually grey).
[FIGURE 6 OMITTED]
In a study of the chromatic induction effect on the Hermann Grid illusion, McCarter (1979) varied the color and saturation of the grids and asked participants to identify the color and the degree of saturation of the ID. McCarter found that participants reported that the color of the ID matched that of the grid and participants had more difficulty perceiving the ID at low saturation levels than they did at high levels. Finally, McCarter reported that the apparent saturation of the ID seen with grids of longer wavelength (yellow and red grids) was higher than that seen with shorter (green and blue) wavelengths.
It is possible that our results differed so markedly from McCarter's results because of two differences in experimental procedure. First, McCarter used a chin rest to limit head movement and maintain a constant viewing distance and angle. She attributed the differences in apparent saturation of the ID for long and short wavelength stimuli to the fixed visual angle and suggested manipulation of visual angle in order to investigate this effect further. Participants in our study were free to vary their head position although we did ask them to try to maintain their viewing distance from the monitor. If the color and saturation of the ID are dependent on viewing angle, we might expect that our participants would not report that the color of the ID matched the color of the stimulus.
Second, McCarter asked participants to match the apparent saturation of the ID using Munsell chips, thereby giving participants the opportunity to use a standard color/saturation stimulus in their assessment of the ID. We did not provide a "standard" by which to judge the color of the ID, and this may account for the tendency of our participants to report that the color of the ID did not match the color of the grid. The majority of our participants reported the ID as being a different color than that of the grid. The most apparent difference can be seen with the results for the yellow grids (Figure 6). Participants were not asked to name the color of the grid, only the color of the ID if they perceived one. In a future study, it would be interesting to have the participants name the color of the grid along with the color of the ID. While gathering data, many participants volunteered that the lowest saturated yellow grid appeared to be green. Therefore, when they identified the ID as green, their responses were coded as a different color than the grid.
Finally, our hypothesis that perception of the ID in yellow grids would be worse than in other colored grids was supported. We found that at middle to high saturation levels, perception of the ID in a yellow grid was significantly worse than in other colored grids. This implies that the visual illusion is perceived differently on a yellow grid than it is on any other color tested. This implication is further supported by the result that gaze location pattern was significantly different with regards to the yellow grids. Perhaps participants were looking directly at the IDs when the grids were yellow and in doing so, were eliminating the perception of the ID more often than with any other color.
Overall, more participants had trouble perceiving the ID on the yellow grids than on any other colored grid. With regards to saturation level, as the saturation increased, participants were less likely to perceive the ID on the yellow grids (Figure 5). One probable explanation for this result is reflectance values.
Reflectance is the amount of light that is reflected by a surface. When yellow is at its most saturated point, it reflects almost the same amount of light at does white (see Figure 7, Omron Corporation, 2007). This implies that participants may have seen a yellow ID, but because the background was white, it was almost impossible to distinguish the ID from the background. A future research opportunity would be to replicate this study and to include grids with other backgrounds, such as black. It would be interesting to see if participants could perceive an ID on the most saturated yellow grid with a black background.
[FIGURE 7 OMITTED]
While we used an eye-tracker, we noticed that participants tended to look at the non-white (colored) part of the Hermann grid array rather than at the white background bars. We also discovered that participants spent less time looking at the more saturated grids than at the lower saturated grids. A possible explanation for this could be a learning curve. The grids were presented to the participants in the same color order for every saturation level and also in order by saturation starting with the lowest level. The participants may have spent more time looking at the lowest saturation grids because they were learning exactly they were supposed to be doing. As they caught on to what we were asking, they needed less time to prepare their answers and so looked at the entire grid for a shorter period of time. Another future research idea would be to present the grids in random order of both color and saturation level to see if participants would still spend less time looking at the higher saturated grids.
Our results are similar to those of Levine et al. (1980). Those researchers found that participants had more trouble perceiving an ID when the intersections were green or yellow. Seeing that participants had trouble with yellow whether it was used as a background color or the color of the squares implies that the reflectance values of the colors used may be a key aspect to how the visual illusion works for the Hermann Grid.
Berbaum, K., & Chung, C. S. (1981). Perceptive field sizes and a new version of the Hermann grid. Perception, 10, 85-89.
Cox, M.J., Ares-Gomez, J.B., Pacey, I.E., Gilchrist, J.M., & Mahalingam, G.T. (2007). Modeling the spatial tuning of the Hermann grid illusion. Spatial Vision, 20(5), 415-436.
DeLafuente, V., & Ruiz, O. (2004). The orientation dependence of the Hermann grid illusion. Experimental Brain Research, 154, 255-260.
Goldstein, E.B. (2010). Sensation and Perception (8th edition). California, Wadsworth Press.
Geier, J., Bernath, L., Hudak, M., and Sera, L. (2008). Straightness as the main factor of the Hermann grid illusion. Perception, 37, 651-665.
Levine, J., Spillman, L., & Wolf, E. (1980). Saturation enhancement in colored Hermann grids varying only in chroma. Vision Research, 20, 307-313.
Lingelbach, B., Block, B., Hatzky, B., & Reisinger, E. (1985). The Hermann grid illusion--retinal or cortical? Perception, 14(1), A7 (Abstract).
McCarter, A. (1979). Chromatic induction effects in the Hermann grid illusion. Perception, 8, 105-114.
Oehler, R., & Spillman, L. (1981). Illusory colour changes in Herman grids varying only in hue. Vision Research, 21, 527-541.
Omron Corporation. (2007). Reflectance of various colors at different wavelengths of light. Retrieved from http://www.ia.omron.com/support/guide/18/further_information.html
Schiller, P. H., & Carvey, C. (2005). The Hermann grid illusion revisited. Perception, 34, 1375-1397. Doi: 10.1068/p5447
Vergeer, M., & van Lier, R. (2010). Capturing lightness between contours. Perception, 39, 1565-1578.
Author info: Correspondence should be sent to: Dr. Barbara Blatchley, Department of Psychology, Agnes Scott College, 141 E. College Ave., Decatur, GA 30030.
Barbara J. Blatchley and Haifa Moses
Agnes Scott College
|Printer friendly Cite/link Email Feedback|
|Author:||Blatchley, Barbara J.; Moses, Haifa|
|Publication:||North American Journal of Psychology|
|Date:||Jun 1, 2012|
|Previous Article:||Pessimistic attributional style and cardiac symptom experiences: self-efficacy as a mediator.|
|Next Article:||The effect of imagery instruction on memory.|