If the visual system splits the image up into separate spatial frequency bands then interactions between channels may have effects on perception. Blakemore and Sutton (1969) found that, after adaptation to a sine wave grating of a particular spatial frequency a grating of a lower spatial frequency appeared to be lower still, and grating of a higher spatial frequency appeared to be of an even higher spatial frequency.
An explanation of the spatial frequency shift requires a model of how activation in a range of channels gives rise to the perception of a specific spatial frequency. One possibility is that spatial frequency is encoded in terms of the distribution of activity across a population of spatial frequency coded channels. Blakemore and Sutton suggested that adaptation to a lower spatial frequency reduces the response in that channel to the test spatial frequency and so shifts the peak of the distribution away from the adapting spatial frequency. Another possibility suggested by DeValois and DeValois is that spatial frequency is encoded in terms of the relative activation of tuned channels. If there are two (A, B) adjacent overlapping spatial frequency tuned channels then if B is more active than A indicates a high spatial frequency and vice versa. If the test grating is chosen to activate A and B equally, then adaptation to a high spatial frequency will reduce the sensitivity of B so that, when the test is replaced, A is now greater than B, which indicates a shift to a lower spatial frequency. These two types of explanation are common in many areas of visual science. One problem with the population code approach is that any pattern will give rise to a distribution of activity across channels. If perception is determined by the most active channel, it is difficult to explain why we see patterns other than gratings.
A similar psychophysical approach can be used to show that we have orientation specific channels. Contrast adaptation is orientation specific as well spatial frequency specific (Blakemore and Campbell, 1969). In addition, an orientation specific aftereffect can be measured. The tilt after effect, which is similar to the spatial frequency shift. Adaptation to a particular orientation has the effect of shifting the apparent orientation of adjacent orientations away from the adapting orientation. These results indicate that channels are tuned to both orientation and spatial frequency.
The fact that adaptation to contrast is orientation tuned, suggests that the site of adaptation is at or beyond striate cortex, which is the first location in the retino-cortical pathway in which orientation tuned cells can be found. DeValois, Albrecht and Thorell (1982) measured the orientation and spatial frequency tuning of V1 cortical cells in the Macaque monkey. They found neurones only responded to a narrow range of orientations and spatial frequencies. Thorell, used 2-deoxyglucose as an indicator of neural activation and found that, when a monkey viewed low spatial frequency gratings, activity was greater near cytochrome oxidase blobs, but when high spatial frequencies were viewed, activity was greatest in the interblob regions. This indicates that spatial frequency channels are ordered from blob regions to inter-blob regions. Orientation channels are organised so that an orderly progression in orientation preference around the blobs can be found.
One can appreciate The information available in separate channels can be appreciated by applying linear spatial filters to real images. The resulting images show the degree of match between the filter and the underlying image structure. By this means we can select out spatial frequency components of images and get a picture of how responsive a particular type of cell would be to a typical pattern. Oriented filters pick out features of the pattern, which have the same orientation and scale. Hubel and Weisel (1977) and Barlow (1972) suggested that simple cells in the visual cortex are feature detectors. We now know that a single cell cannot unambiguously tell us whether there is a feature of a particular type in the image. This is because a non-optimal high contrast stimulus will drive the cell as well as an optimal low contrast stimulus. Alternatively, we can think of simple cells in visual cortex as filtering the image.
The visual system does not perform a Fourier analysis on the visual scene because a Fourier analysis properly applies to the whole image and neural receptive fields in V1 are small and local. Spatial frequency selective filters may play a role in encoding periodic textures but again simple cells have small receptive fields relative to the scale of many natural textures. It is likely that the spatial filters provide a much richer and more useful representation of the image than a point by point representation of image brightness. Analysis is regional rather than pointwise and this means, for example, we can faithfully represent the complete signal on the retina even though it is only sampled at a number of points by retinal receptors or neuronal receptive fields. Its usefulness however needs to be defined relative to the operation of mechanisms that use this representation as an input and compute properties of the world such as surface shape, object motion and surface depth.
This experiment will aim to reproduce the findings of Blakmore and Sutton with improved methodology. The hypothesis is that the longer the period of adaptation the higher the participants PSE. This means that the PSE should be higher for experiment 1 and for experiment 3 where no adaptation took place the PSE mean for the participants should be around 100.
Method
Design
The design employed was a one-way between subjects ANOVA. The between-subjects factor has three different levels, experiment 1, 2 and 3. The independent variables are the three experiments used which all have different adaptation durations. The dependant variable is the PSE.
Participants
The participants consisted of 87 Psychology students who were divided into three groups each of which did one experiment. 29 paticipants did experiment 1 and 3 and 30 participants did experiment 2. The age of the participants ranged from 19 to 41 and comprised of approximately 80% females and 20% males.
Materials
A computer program was used to carry out the experiment. All the experiments used the same patterns. The stimulus patch measured 8 degrees (measurement of visual angle) horizontally and 12 degrees vertically. The stimulus patch was divided horizontally by a horizontal fixation line, which was centred within the patch and measured 2 degrees long. The gratings were presented above and below the fixation line and measured 10 degrees horizontally, by 6 degrees vertically. It is important the participant does not shift his gaze from this fixation line. The bars of each grating were vertical. The grating elements were specified by their spatial frequency in cycles/deg. The adaptation stimulus was half the spatial frequency of the test stimulus which means the adaptation occurred at one octave less than the test stimulus.
Procedure
The class was divided into three groups. Each group took part in one experiment. The computer program was designed to enable the participants to select the experiment they were going to carry out. The adapting stimulus was presented for either 20 seconds (exp. 1), 5 seconds (exp. 2), or 0 seconds (exp. 3). Following adaptation there was a short delay of 0.5 seconds before the test and reference pattern appeared for 1second. In experiment 3 only the test and reference patterns were presented but the duration of the presentation will be the same as for the other two experiments. For each experiment, 10 presentations of each reference stimulus were made totalling 70 presentations. The reference stimuli were presented in a random order. The participants were asked to make the judgement of whether the bottom pattern was coarser or finer than the top pattern by responding accordingly. Once the experiment had ended, the participants were asked to record their PSE. The data was recorded for analysis.
Results
The raw data comprised of the participants point of subject equality measurements (PSE). The PSE was measured in terms of the percentage of the test spatial frequency.
The data, as expected revealed that the PSE mean was lower for experiment 3 than for experiment 1 and 2. There was however no significant difference between the mean for experiment 1 and 2 as would be expected if adaptation duration has an effect on the PSE obtained. This will be looked at in the discussion.
The descriptive statistics (see appendix) show that the standard deviation for the participant’s scores in experiment 3 are lower than for the other two experiments. This means there is less difference between the participant’s scores in experiment 3; this is because no adaptation took place therefore the participants would have all responded in a similar way.
The ANOVA revealed there was a significant effect of experiment. (F (2,85) = 16.926, p<0.05. This means that there was a significant difference between the experimental conditions.
However, Levene’s statistic is significant revealing that the assumption of homogeneity of variances has been violated and therefore the Games-Howell post hoc tests are used.
Employing the Games-Howell post hoc test significant differences were found between experiment 1 and experiment 3 (p<0.05), and between experiment 2 and 3. There was no significant difference between experiment 1 and 2 (p=1.4097). This suggests that it makes no significant difference to the results how long the adaptation duration period is. 20-second adaptation duration made little difference when compared to the results of experiment 2, which had 5 sec duration. This will be looked at further in the discussion.
The graph shows the effect the different experiments had on the participants mean PSE scores.
Discussion
The main results showed that there was a significant effect of the type of experiment used in predicting the PSE. For experiment 1 the adaptation duration was 20 seconds and for experiment 2 the adaptation duration was 5 seconds. Surprisingly the PSE mean was not that much lower for experiment 2 when compared to experiment 1. This may be because it does not take as long as 20 seconds for the adaptation to take place and 5 seconds is a sufficient amount of time for the same effect to happen. As expected the standard deviation between the participants scores for experiment 3 was much lower than for the other two experiments. This is because no adaptation took place and the participants had nothing affecting their response to the stimulus. These results back up those found by Blakemore and Sutton as adapting to a spatial frequency effects the way another spatial frequency is perceived.
There are a number of possible ways to improve this research. Firstly better control over confounding variables could be put into place. For example if the experiment is set in stricter surroundings increased concentration will be achieved and more accurate results. As mentioned it was imperative that the participants focus there gaze at the fixation line while completing the experiment. Quieter more controlled surroundings would have minimised the risk that the participants gaze may shift if distracted. An alternative suggestion would be to use more adaptation times (ie, more experiments) ranging from 1 second to 20 to establish exactly how long is needed to effect the participants responses sufficiently. Other types of patterns could also be used to see if the same effect occurs. It may also be more valid to enable all participants to take part in all the experiments and compare the results to account for individual differences.
References
Blakemore, C. & Campbell, F.W. (1969). On the existence of neurones in the human visual system selectively sensitive to the orientation and size of retinal images. Journal of Physiology (London), 203, 237-260.
Blakemore, C. and Sutton, P. (1969) Size adaptation: A new aftereffect. Science, 166: 245-247
Campbell, F.W. & Robson, J.G. (1968). Applications of Fourier analysis to the visibility of gratings. Journal of Physiology, 197, 551-566.
De Valois, R. L. and De Valois, K. K. (1990) Spatial Vision, Oxford University Press.
Goldstein, E.B. (1999) Sensation and Perception 5th ed. Brooks/Cole Publishing