Adaptation in Blind Walking

1

Results

In this study subjects were tested to see if they would adapt during a blind walking task. If this adaptation were to occurit would decrease the subjects tendency to underestimate distances, or in other words, subjects would walk a greater distance while trying to estimate the distance to a target after, when compared to before, adaptation has occurred. It was hypothesized that the mean of the distance walked in the non-visual condition would be higher then that of the visual condition. As well the mean of the distance walked in block 3 would be higher then that of block 1.

The percentage of error from the target was calculated for all conditions. The mean of the percent error for the visual condition was -0.015 with a standard deviation of 0.1833. The mean of the percent error for the non-visual group was -0.0338 with a standard deviation of 0.1719 (Figure 1). The means of the percent error for the block 1, 2 and 3 were respectively 0.0038, -0.0528 and -0.0242 with standard deviations of 0.207, 0.140, 0176 (Figure 2).

A 2 (conditions) X 3 (blocks) repeated measures ANOVA showed that there was no significant main effect for the twoconditions (F(1,31) = 0.198, p > 0.05) and for the three blocks (F(2,62) = 2.074, p > 0.05). As well it was found that there was no significant interaction effect (F(2,62) =.953, p > 0.05)(Table 1).

Discussion

The purpose of this experiment was to determine if prior exposure to blind walking would lead to adaptation. If this adaptation were to occur it would increase a subject’s perception of how great a particular distance is. No significant results to support this hypothesis were found in this study. Prior exposure to blind walking had no significant effect on distance estimation; as well the number off trials had no significant effect. Had there been a significant effect for either of these two conditions, the blind walking task would be an inaccurate test for determining accuracy without visual input. The reason for this would be that when adaptation occurs it, subconsciously, alters a person perception of distance, therefore leading to greater distance estimations. This would then lead to greater variability in results during the blind walking task, as well improved accuracy could be explained by adaptation instead of learning. However this appears to not be the case. Our results do not show any adaptation occurring during our experiment as there were no significant effects (Table 1).

Elliot (1987) noticed a trend in the results from two different experiments, Elliot (1986) and Elliot (1987), in which the greater amount of practice a subject received, the greater their tendency to underestimate decreased during the experiment. This trend was unexpected and due to the methodology of the experiment, unexplainable. Testing non-visual against visual practice and early trial blocks against later trial blocks was this experiment’s attempt to explain Elliot’s (1987) observation. However due to the fact that this experiment failed to find any significant results,adaptation does not appear to be the cause of this trend. In a different study, Proffitt (2003) did find significant results when testing for adaptation. Proffitt (2003) did not use a blind walking task in his experiment though. Instead Proffitt (2003) had subjects verbally estimate distances, before and after, either a non-visual or a visual adaptation task. The different findings in these two studies could possibly be explained by errors in our methodology.

There are several possible explanations for why the results in this study did not match the hypothesis. First, this experiment used a very small sample size, eight subjects. A small sample size increases the probability of error. By increasing the amount of subjects in the experiment, a much more accurate result could possibly be obtained. As well, when the amount of subjects is increased, the study would be a better representation of the population as a whole.To decrease the amount of variation in the study caused by a small sample size a within subjects test was used. However this only corrects for a limited amount of variance. As well, technically nine subjects participated in the study, as one subject had to drop out after only competing one of the conditions. The experimenters running these particular trials did not put this new subject through both conditions, instead this new subject participated in only one condition which was then compared to the one other subject’s performance in the opposite trial. This completely eliminates the benefit of using a within subjects design.

Secondly, the subjects that participated in the experiment also helped design the experiment. This could greatly bias their actions during the experiment as they know exactly what the experiment is looking for. Typically it would be expected that this problem would lead to the subjects wanting to perform better and therefore walking longer on later trials. This does not appear to be the case in this experiment, as rather then walking longer the subjects walked shorter distances as the number of trials increased. However knowing what the experimenters, or more accurately, thinking you know what the experimenters are looking for leads a subject to perform differently then they would normally.

The observation that subjects tend to walk shorter distances on later trials can be explained by the third possible problem with the methodology, lack of motivation. Even though subjects were experimenters as well, there was no motivation to really focus on the task. After talking with the subjects shortly after the experiment was completed, it appeared as though they were just going through the motions of, and not actually focusing on, the experiment. This general lack of caring could have led the subjects to make no attempt to improve during the task, or to even make an attempt to remember, during a trial, the distance they were expected to walk. This problem could easily be solved by offering some sort of compensation for the subject’s time or by offering a prize to the most accurate subject. By doing either of these two things you are giving subjects a reason for trying to reach their target accurately. Unmotivated subjects can greatly skew the results because it does not give an accurate representation of their true abilities.

Lastly there was a problem with the equipment used. The blindfold that was used during the task did not fully block the vision of the subjects. Rather it allowed for some sight out of the bottom. To compensate for this problem subjects were asked to not look down, which allowed them to see the ground; however there was no way to confirm that subjects followed these instructions. This problem could explain why no adaptation took place during the blind walking task, as subjects were never truly blind.

If this experiment were ever replicated there are several simple ways by which the method may be improved. First, and foremost, the subjects should not be the experimenters as well. Secondly subjects need some sort of motivation to do well and focus on the task at hand during the experiment. Thirdly a greater number of subjects should be used to increase the accuracy of the experiment. Lastly, faulty equipment should not be used, especially equipment as important as a blindfold in a blind walking task.

References

Elliot, D. (1987). The Influence of Walking Speed and Prior Practice on Locomotor Distance Estimation.

Journal of Motor Behavior, 19(4), 476-485

Proffitt, D.R., Stefanucci, J., Banton, T., & Epstein, W. (2003). The Role of Effort in Perceiving Distances.

Psychological Science, 14(2), 106-112

Appendix

Tests of Within-Subjects Effects
Measure:MEASURE_1
Source / Type III Sum of Squares / df / Mean Square / F / Sig.
condition / Sphericity Assumed / .998 / 1 / .998 / .198 / .660
Greenhouse-Geisser / .998 / 1.000 / .998 / .198 / .660
Huynh-Feldt / .998 / 1.000 / .998 / .198 / .660
Lower-bound / .998 / 1.000 / .998 / .198 / .660
Error(condition) / Sphericity Assumed / 156.580 / 31 / 5.051
Greenhouse-Geisser / 156.580 / 31.000 / 5.051
Huynh-Feldt / 156.580 / 31.000 / 5.051
Lower-bound / 156.580 / 31.000 / 5.051
block / Sphericity Assumed / 11.559 / 2 / 5.780 / 2.074 / .134
Greenhouse-Geisser / 11.559 / 1.817 / 6.361 / 2.074 / .139
Huynh-Feldt / 11.559 / 1.924 / 6.008 / 2.074 / .136
Lower-bound / 11.559 / 1.000 / 11.559 / 2.074 / .160
Error(block) / Sphericity Assumed / 172.808 / 62 / 2.787
Greenhouse-Geisser / 172.808 / 56.334 / 3.068
Huynh-Feldt / 172.808 / 59.648 / 2.897
Lower-bound / 172.808 / 31.000 / 5.574
condition * block / Sphericity Assumed / 3.227 / 2 / 1.614 / .953 / .391
Greenhouse-Geisser / 3.227 / 1.935 / 1.668 / .953 / .389
Huynh-Feldt / 3.227 / 2.000 / 1.614 / .953 / .391
Lower-bound / 3.227 / 1.000 / 3.227 / .953 / .337
Error(condition*block) / Sphericity Assumed / 104.985 / 62 / 1.693
Greenhouse-Geisser / 104.985 / 59.973 / 1.751
Huynh-Feldt / 104.985 / 62.000 / 1.693
Lower-bound / 104.985 / 31.000 / 3.387

Table 1. Repeated measures ANOVA for within-subjects effects showing no significant results for any comparisons.

Figure 1. Percent error versus distance showing the means for the non-visual and visual conditions.

Figure 2. Percent error versus block number showing the means for both non-visual and visual conditions