Abstract
The present experiment tested the automatic activation against the non-automatic activation hypothesis by altering the angular orientation of characters in a mental rotation task. Two hypotheses were involved. One, the mean reaction time would not change due to the angular orientation of characters. Two, the mean reaction time for large and unusual orientation will be longer than the reaction time for small and usual ones. Participants engaged in mental rotation task to decide if stimuli were presented the right way round or mirror image of the target figure. The mean reaction times for large orientations were longer than small ones. These results above conformed to Hypothesis 2, but disagreed with Hypothesis 1.
Introduction
This paper was concerned with mental rotation in the role of object recognition. This whole concept of mental rotation was based on mental imagery, which was a mental picture represented internally in a person’s mind. The topic of mental imagery, mental rotation and object recognition were researched by numerous psychologists over centuries.
The idea of mental imagery originated from the theories proposed by Kosslyn and Thompson (2003). They suggested that, in the perspective of Perceptual Anticipation theory, mental images were created when one looked forward to perceiving an object so eagerly that the image of the stimulus was generated in the early visual cortex. Visual memories of shapes were stored in an abstract code in the long-term memory initially, and then were made explicit by producing patterns in the early visual cortex. Correspondingly, in Propositional theory, mental images were language descriptions rather than visual images.
Mental imagery was further explored by Shepard and Klun (1973, described in Cooper and Shepard, 1973), when subjects were requested to identify rotated alphanumeric characters as “normal” or “backward” (mirror-image). Some subjects were informed the identity and orientation of test stimulus beforehand. These subjects required considerably less time to make a decision than those not informed. They concluded that subjects initiated an image of the character internally, and then rotated it to an appropriate orientation, which allowed them to compare the mental image with the visual test character. This study implied that mental imagery does exist for mental rotation to be carried out.
Additionally, the effect of mental imagery to the cerebral cortex was investigated (Roland & Friberg, 1985; cited in Farah, 1989). They compared the regional cerebral blood flow when subjects stayed at rest and during the visual imagery task. They discovered that there was a massive blood flow to the posterior regions of the brain, which was the place responsible for higher level of visual processing.
Moving on mental rotation, Shepard and Metzler (1971) involved subjects presented with pairs of perspective line drawings. Their task was to determine whether the two drawings of cube complexes were congruent with the three-dimensional shape. Results showed that the overall means of reaction times at different angular orientations for picture-plane pairs were higher than depth pairs. It was found that mean reaction time was higher for large angular rotation than small ones as well.
Moreover, the effect of head tilting on mental rotation was examined (Corballis, Zbrodoff & Roldan, 1976). Subjects were instructed to decide whether the characters presented on cards was “normal” or “backward” in three conditions: head upright, head tilted left or head tilted right. Results reflected the fact that judgments of “normal” were significantly faster than judgments of “backward”. It also conveyed that judgments made with head tilted right took the longest response time, followed by head tilted left and then head in upright position. In addition to the above results, they also found that judgments about the stimuli ‘R’ and ‘7’ were the fastest.
In 1978, Corballis, Zbrodoff and Roldan carried out another experiment to observe the effect of head tilting on mental rotation. Instead of displaying alphanumerical stimuli, a pattern or symbol was presented in different angular rotations. They were also to decide whether the rotated versions of the symbol were “normal” or “backward” in the three head tilting conditions. Opposing to the results above, the mean response time overall was the highest when tilted left, then head upright and at last when head tilted right. This showed that the test stimuli presented played a critical role in mental rotation.
Furthermore, Cooper and Shepard (1973) conducted a study on the time required to prepare for a rotated stimulus, which included participants to respond to each test stimulus on whether they were “normal” or “backward”. Participants were tested under the “B-condition” (both the identity and the orientation were provided before the actual rotated test character) or “N-condition” (no information regarding either the upcoming stimuli or orientation was provided). In the “B-condition”, it was found that when the orientation cue persisted for a longer period, the mean reaction times taken were shorter; while in the “N-condition”, the mean response time took the longest at 180 degrees.
Although the above studies provided insight to mental rotation, few problems were raised. There were at most 12 subjects involved in each of Shepard and Metzler (1971) and Corballis, Zbrodoff & Roldan’s (1976, 1978) experiment. The reliability of these experiments was low due to the small number of participants. Besides, the high ratio of female to male participants would decrease the reliability as well. This effect on mental rotation will be further explored in the discussion.
This experiment was a replication of the “N-condition” of Cooper and Shepard’s (1973) study as there was no advance information concerning the identity or orientation of the upcoming test stimuli provided. The main aim of this study was to test the automatic activation against the non-automatic hypothesis of object recognition.
As an induction, a simple theoretical model of object recognition was described. Firstly, the visual system was involved to process the appearance of the external stimulus and a memory of it was stored in the memory stores. When the external stimulus is shown again, the memory of the previously learnt object is retrieved from the memory stores. The automatic activation hypothesis (AAH) advocated that object recognition occurs automatically in any presentation, while the non-automatic activation hypothesis (NAAH) suggested that automatic recognition only happens in previously learnt objects.
Participants were involved in a mental rotation task in this experiment. By making use of the purpose-written program on lab-computers, they were instructed to determine whether the stimulus was presented in right way round or mirror image of the target figure. A mental rotation task was involved because it is a good visual task to measure participants’ response times for stimuli displayed at various orientations and letters and numbers were used as test stimuli because they both have a fixed orientation that we are familiar with.
It was predicted that the mean reaction time would not change due to the angular rotation of characters (Hypothesis 1) in accordance with the AAH we proposed. It was also expected that the mean reaction time for large and unusual orientation of characters would be longer than that for small and usual orientations (Hypothesis 2).
Methodology
Participants
A total of 59 subjects (2 excluded due to their anomalous nature), all first year Applied Psychology undergraduates at Queen’s Campus, participated the experiment. Among these subjects, there were 9 males and 48 females, aged from 18 to 41 years (mean age = 19.53, SD = 4.40). 13 of them were non-native speakers.
Apparatus
Lab-computers with purpose-written program were used in the experiment. Standard and mirror image versions of letters and numbers (G, J, R, 2, 5, 7) were involved as the test stimuli. A result record sheet was used to note details such as gender, age and performance of subjects. By the performance of subjects, the number of correct responses made at each orientation, the reaction time taken to respond and the overall accuracy of subject was documented.
Design
This experiment had a within-subject design in which all 59 participants underwent experimental trials for the six orientations. The independent variable of this experiment was six orientations, which consisted of 0, 60, 120, 180, 240 and 300 degrees. While the dependent variable was the time period participants required to decide whether testing stimuli was standard or mirror-image of the target figure. The orientations and test stimuli were presented in randomized order as the participants were given an overview of stimuli and orientations involved. The whole experiment was composed of two practice trials and 144 experimental trials, which were then broken down into 12 blocks with each block of 12 trials. Subjects were required to repeat the practice trials if they made more than two errors, in order to obtain an accurate reaction time for each orientation and character by making sure that subjects understood the instruction of the task. Also, data acquired from subjects with overall accuracy lower than 60% were discarded, this is to maintain the accuracy of the mean and standard deviation, as those data were anomalies.
Procedure
All participants worked individually and their task was to indicate which way the character is facing. Before the trials, participants were informed what stimuli and orientations are involved. Participants made use of the purpose-written program on the computers and underwent practice trials as well as experimental trials. Each trial was arranged in blocks of 12. If participants made more than two errors in the practice trials, they were required to retake the trials. Each block was initiated by pressing the space bar and characters were presented sequentially in one of the six orientations and either the right way round or mirror reversed. Participants were asked to press one of the two keys, ‘M’ (right way round) and ‘Z’ (mirror reversed), as soon as they can after the character was presented. Between each trial, there was an interval of 500 milliseconds. After participants completed the block, they had to record the results on the record sheet.
Results
The response times for participants to make a decision whether the testing stimuli were standard or mirror image of the target figure were analysed. Table 1 presented the mean and the standard deviation of the subjects’ response times for each of the six orientations, while Figure 1 displayed mean reaction time as a function of orientation of stimuli.
Table 1. Mean and standard deviation of subjects’ response times (N=57)
Degrees of orientation
0 º 60 º 120 º 180 º 240 º 300 º
Mean 1154.6491
1465.8246
1576.2982
2356.1579
1771.8596
1387.6667
(SD) (301.4980) (703.4066) (613.0200) (1032.1342) (671.1342) (422.7720)
Table 1 showed the mean and standard deviation of subjects’ response times increased at greater degrees of orientation. The mean response time was the highest at 180 degrees (2356.1579), followed by 240 (1771.8596) and 120 deg (1576.2982), while the largest standard deviation occurred at 180 deg (1032.1342), then 240 (671.1342) and 120 deg (613.0200) as well. Both of these supported the hypothesis that large and unusual orientation of characters will result in longer response times than small and usual ones.
Figure 1. Line graph showing mean reaction times at six different angle rotations of characters. (The point plotted at 360 degrees duplicates the point at 0 degrees)
Figure 1 presented the mean reaction time in relation to the six angle rotations of characters clockwise from the target figure. All the points plotted in Figure 1 were independent, except for the point at 360 deg, which wass the duplicate of the point for 0 deg. The results in figure 1 illustrated that subjects took the longest time to decide whether the test stimulus was right way round or mirror image of the target figure at 180 deg and they took the shortest at 0 deg. Overall, the shape of the graph was non-symmetrical and the mean reaction time at 240 degrees (1771.8596) was greater than that at 120 degrees (1576.2982), while the mean reaction time at 300 deg (1387.6667) was smaller than 60 deg (1465.8246).
Discussion
The main aim of this study to test the automatic activation against the non-automatic hypothesis of object recognition was achieved. From Table 1 and Figure 1, it was illustrated that participants took longer to respond at large rotations such as 180 and 240 deg. This above result confirmed Hypothesis 2, where it was stated that the mean response times for large and unusual orientation were greater than the small and usual ones. However, these results disagreed with Hypothesis 1, as the mean response times did alter due to the orientation of characters.
General results shown in Figure 1 agreed with Cooper and Shepard’s (1973) study, but there were few parts that contradicted. To begin with, the reaction time in their study was a “sharply increasing function from the standard upright orientation” (cited in Cooper & Shepard, 1973). Figure 1, although exhibited an increasing function from 0 to 180 degree, the increase was quite weak in comparison.
In addition, the mean reaction as a function of orientation of the test stimulus obtained in the above study had a “remarkably symmetrical shape of function about 180 deg” (cited in Cooper & Shepard, 1973). On the contrary, the result obtained in Figure 1 was non-symmetrical. This was suggesting that, in reality, the increase in reaction time for both clockwise and counterclockwise rotations were not equal.
Table 1 displayed that the standard deviation of mean reaction time at 180 degrees was the largest. This could be due to the differences in strategies (Liesefeld & Zimmer, 2010), where some participants flipped alphanumerical stimuli upside down and thus resulted in a shorter reaction time. This implied the limitation of this experiment – inability to control strategies, and this factor might have contributed to the great standard deviations at 180, 240 and 120 degrees.
Apart from that, there were few confounding variables existed in this study. First, there were individual differences, for instance, gender and age. It was revealed that, in terms of gender, males have better performance in mental rotation tasks than females (Janssen & Geiser, 2009; Voyer & Jansen, 2016). Similarly, it was found that children and older adults took longer time to respond than adults (Kaltner & Jansen, 2016). Relating this to our study, these factors could have contributed in the unusual results, as there were more female participants than males and the age distribution was concentrated at 19 years. Therefore, future studies should engage equal numbers of male and female subjects and recruit subjects from different age groups.
Another confounding variable to be controlled was whether female participants take hormonal contraceptives. In accordance with Grisksiene and Ruksenas (2011), the mean reaction times of hormonal contraceptive users were higher than non-users. This factor may be accountable for long response times at 180 and 240 degrees, as there were more females than males in the experiment.
Considering the above findings, we tested the AAH against the NAAH and tried to achieve a better understanding on how the angular orientations of test stimuli affect metal rotation. To conclude, object recognition happens automatically only in previously learnt objects and the mean reaction time for large and unusual orientations will be longer than the reaction time for small and usual ones.