One of the classical debates in psychology is to what extent our ability to perceive the world is biologically-predetermined versus the product of postnatal experience. Infant perception cannot be measured directly, but indirect measures such as physiological and behavioural ones are possible. Perception of depth has been investigated by the “visual cliff” paradigm - first used by Gibson and Walk . Infants of crawling age, around 7 months upward, were placed on a dark surface next to a transparent sheet of safety glass at a height of about 1 m above the floor, and were incentivised by their mothers to crawl across it. Only very few did, but when put on the transparent safety glass, they happily crawled over, which leads to the assumption that these infants could accurately interpret the depth cues. Even younger infants showed a change in heart rate, when held over the different parts in the experimental setup of the “visual cliff”.  At this age, they can already distinguish different depths, but only learn to appreciate fully what they “mean” once they gain locomotor experience. Hence, human infants seem to learn which cues in the visual input signal depth. The question we are addressing here is whether this ability to learn new cues to depth is still present in human adults? Backus et al.  have shown that adults can learn to associate any arbitrary stimulus feature with depth and that these features by themselves would elicit depth perception. In this experimental design, our aim is to confirm that vision in adults is still modifiable by association training of specific retinal locations with certain depth arrangements of stimuli, such that the presentation of a depth-ambiguous stimulus at a particular location would cause it to be perceived as having a certain depth arrangement.
Participants were trained to temporarily adopt to a location (above or below fixation) as a depth cue by performing a depth discrimination task for 3 blocks of 60 trials consisting of a fixation cross (500ms), which was replaced by the Question “Which side is nearer/further?” for 1 second and then by a fixation cross for a further 500ms followed by a rectangle in one of two positions, either above or below fixation.1 If it appeared above fixation, binocular disparity would signal either that the left-hand side was nearer than the right (Group A), or the right-hand side nearer than the left (Group B).2 In both groups, the respective opposite depth arrangement was the case when the bar appeared below fixation.
In the test block of trials, the rectangle was presented to just one eye 1The experiment was conducted using iMac computers with a 17” screen and running Psyscope Experimenter Generator Software. 2Due to a lack of observers having reported depth perception in these training trials, group A and B were collapsed into a single group in the following. on 40 of the 60 trials3, yielding no binocular disparity and, therefore, was ambiguous in terms of its depth arrangement. For these ambiguous stimuli, it was predicted that subjects would perceive the stimulus to have the same depth arrangement as previously observed in the respective location. Two paired one-tailed samples t-test were performed
- one based on the H 0 ,training hypothesis that there is no significant difference in the number of correct and incorrect answers during the training session, and furthermore, one based on the H 0 ,test hypothesis that there is no significant difference between the number of trials, in which a test block was perceived same or opposite to the previous training stimuli at the respective same location as in the training trials
- in a group consisting of 14 undergraduate students at Cambridge University, that reported having experienced depth perception, using a critical value of α = 0 . 05.
We have shown that the group of 14 undergraduate observers that have reported seeing depth were indeed able to report the depth arrange- ment of stimuli presented in the training blocks. In both a one-tailed and two-tailed paired t-test (see page 5), we found t (13)0 . 05 = 8 . 18, and thus we reject H 0 ,training in both cases: There is a significant dif- ference between the number of correct and incorrect answers during the training session, clearly shifted to the correct answers being more salient, hence these observers as a group could indeed reliably report the depth arrangement of stimuli presented in the training blocks. One clear exception is subject 13, but even by including him or her, we have to reject H 0 ,training in the training trials and retain H 0 ,test for our actual experiment, in which observers did not perceive, more than would be expected by chance, the same depth arrangement for an ambiguous stimulus at a particular location in the test block as they had perceived for the unambiguous stimuli during the training blocks. Even by exclu- sion of subject 13, we still need to retain H 0, which states that there is no significant difference between the number of trials, in which a test block was perceived same or opposite to the training stimuli at the same location, given that t (13)0 . 05 = 0 . 2.
First of all, there was a a two-alternative forced choice in the test block, hence no response was available that signalled “no clear depth percept”, which might have led to guessed answers. It might have been helpful to exclude these by including a question such as “Please indicate on this scale from 1 to 5 how certain you are about your answer.” All 3The other 20 trials of the test block consisted of unambiguous depth stimuli and were thus not included in the analysis. answers below 2 could then be excluded. Alternatively, an additional answer possibility “Don’t know” could be introduced, so that only proper “Incorrect” and “Correct” answers would be taken into account. Secondly, the limitations of the experiment have been left out: The experimental design assumed that observers would readily perceive depth in the training trial stimuli. However, only 14 observers reported to actually have seen depth in the training trials. This corresponds to 10% of the subjects, so we would need to redesign our experiment in order to test whether vision is still modifiable in all adults. Thirdly, many subjects reported that they had difficulties due to the constant switch in the question “Which location is nearer/further?”. This might have led to confusions by trying to act as accurately as and in the shortest reaction time possible.
In conclusion, we verified that the 14 subjects could indeed reliably report depth as a group, but this did not lead onto the finding we expected based on Backus’ experiments 2: that observers would report perceiving, more than would be expected by chance, the same depth arrangement for an ambiguous stimulus at a particular location in the test block as they had for the unambiguous stimuli at that location in the training blocks. Hence, we rejected the first H 0 ,training hypothesis that there is no significant difference in the number of correct and incorrect answers during the training session, and furthermore, retained the second H 0 ,test hypothesis that there is no significant difference between the number of trials, in which a test block was perceived same or opposite to the training stimuli at the same location.
1The experiment was conducted using iMac computers with a 17” screen and running Psyscope Experimenter Generator Software.
2Due to a lack of observers having reported depth perception in these training trials, group A and B were collapsed into a single group in the following.
3The other 20 trials of the test block consisted of unambiguous depth stimuli and were thus not included in the analysis.
- Quote paper
- Laura Imperatori (Author), 2014, Sensory Motor Development and Plasticity, Munich, GRIN Verlag, https://www.grin.com/document/276479