Virtual Reality Monitoring: How Real is Virtual Reality?

by Keith Hullfish

CHAPTER 4 COGNITIVE EFFORT AS AN ARTIFACT

4.1 In Source Monitoring

In the first experiment, cognitive effort was manipulated by varying the number of external cues specifying the response. Subjects were presented with a list of words to which they either had to listen to the experimenter respond or they had to generate a response themselves. Responses had to be an instance of the category defined by the presented word. On a response sheet, subjects were either cued with the first letter of the appropriate response or received no aid. When cued with the first letter, it was expected that subjects would require less effort to generate a response which would cause greater confusion between self-generated and perceived (i.e. presented by the experimenter) memories.

After completing the list, an unexpected test was given in which the experimenter read a word and subjects had to identify its origin. The results confirmed one of their original hypotheses that evidence of cognitive effort did provide a cue to source. Subjects correctly discriminated origins more frequently when there was no first-letter cue for self-generated responses.

Cognitive effort also appeared to be characteristic of memories of self-generated responses. Even as these responses were more confused with perceived responses in the first-letter condition, they were still more discriminable than the latter. In this condition origins were the most similar perceptually. If monitoring decisions were based solely on perceptual qualities, then each origin should have been equally discriminable. Since any additional criteria would tend to reflect how the memory was established, it is likely that the effort needed to generate responses characterizes memories of these imagined events.

Similar patterns were found in another experiment, however, the method with which automaticity depended on the external cues was changed. In the second experiment, the difficulty did not rely on the similarity with external cues, but was specified by the typicalness of the possible responses. The task was the same but the perceived responses were written and all self-generated responses were constrained by a first letter cue. The response was either an opposite or a category instance of the presented word. For the latter, the first-letter constraint limited responses to either high or low frequency category instances. Presumably, low-frequency category instances are less typical and more difficult to generate. Even after a 10-day retention interval, these self-generated responses were, as predicted, more identifiable. Since self-generated responses did not differ perceptually, these decisions must be based on the relatively more cognitive effort needed to generate them. This supports the assumption made in the first experiment that manipulating the number of external cues also manipulates the degree of cognitive effort in specifying a response.

Of particular note in the last experiment was the confidence subjects had in their attributions. Memories of self-generated responses had a higher confidence as compared to perceived responses. Furthermore, subjects were more confident in their correct identifications than when they confused origins. This confirmed expectations that confidence is also a sensitive measure of origin. This discrepancy was greatest for low-frequency category responses which indicates that cognitive effort is characteristic of memories of self-generated responses.

A study by Finke et al. (1988) used spatial images instead of words in exploring the role of cognitive effort in reality monitoring. If images are difficult to form, then their memories should be more distinguishable when identifying origin. Under a cover task of rating their complexity, subjects were presented with horizontally or vertically symmetrical patterns. In the imagined condition, they were presented with half of the pattern, divided about its axis of symmetry. Some subjects were instructed to imagine the complete form based on this symmetry while others served as a control and only had to rate the forms. From an initial experiment, it was found that horizontally symmetrical patterns were rated more difficult to complete. Given the original premise, it was expected that memories of imagined vertically-symmetrical forms would be more confusable with forms that were perceived as whole.

A surprise recognition test was then given consisting of whole versions of the forms consistent with their original orientation. It was found that discrimination between half and whole forms was poorer when subjects were instructed to imagine a complete vertically symmetrical form as compared to the control. As predicted, the lack of evidence for cognitive effort in these memories led subjects to confuse them with memories of perceived forms. Since whole patterns remained unchanged, such effort would presumably be a characteristic of memories of imagined two-dimensional patterns.

4.2 Its Potential in Virtual Reality

First, naturalness may reflect the match between the physical environment and people's representation systems (Slater & Usoh, 1993). If its technical limitations prevented the virtual environment from matching these expectations, then subjects may be required to elaborate more with a priori knowledge in order to generate an equivalent experience. For instance, reading letters on a low resolution display may require more effort to search for the correct interpretation if these letters are fuzzy. Or people may prefer to represent their experience either visually, acoustically, or kinesthetically (Slater & Usoh, 1993). A person who prefers to represent their environment acoustically might have more difficulty transposing predominantly visual environment to understandable information as compared to someone who is more visually sensitive. These are examples of cognitive effort as an artifact of omission.

Cognitive effort may also be an artifact of commission. Naturalness has been sought in the interaction with the environment (Wells, 1992; Lavroff, 1992; Wickens, 1992; Zeltzer, 1992). This can be achieved through forming a closed-loop where afferent and efferent signals match (Ellis, 1991; Loomis, 1992; Sheridan, 1992; Held &apm; Durlach, 1992). For instance, in acquiring and retaining a pattern of spatially-distributed objects, it would be expected that a certain amount of afferent signal would always be seen while moving your normal field-of-view over the field-of-regard. If the smooth flow of signals is mismatched or disrupted due to a restricted field-of-view, effort may be needed to adapt to these new constraints. In the example above, if the expected afferent signals are not received, one may need to retain more information in working memory in order to integrate over the positions of objects that would have normally been seen in the same field-of-view. Thus, this restriction would generate undue effort in the process of committing the same act.

There is some empirical evidence to support the existence of cognitive effort as an artifact of commission in experiencing virtual reality. In particular, a limited field-of-view may cause cognitive effort in integrating over spatially-distributed information. Wells and Venturino (1989) measured this difficulty using performance on a dual task consisting of maintaining the position of an inverted "T" against the disturbance of forcing functions. Performance on this task, as measured by tracking error, suffered with decreasing field-of-view while subjects also monitored and shot at targets. This suggests that more cognitive effort was needed to perform the first task when field-of-view was smaller.

Sholl (1993) also used a dual task to study the effect of a restricted field-of-view in developing survey knowledge during navigation in a large-scale real environment. The study found that the limited field-of-view of 5¡ inhibited the development of survey knowledge. Mean absolute pointing error to target as a function of the number of turns made in the environment was the dependent variable. It was expected that increasing the number of turns would decrease pointing accuracy if survey knowledge was not acquired. Control subjects who had no visual restriction and did not count backwards while navigating between targets did develop survey knowledge. However, subjects with a restricted field-of-view and those with a full field-of-view and dual task both showed a linear trends in their pointing error, indicating that only route knowledge was formed in these conditions. Given the premise that survey knowledge is not formed automatically, but requires cognitive effort, these results indicate that both the dual task and restricted field-of-view required cognitive resources that would have been needed to develop survey knowledge. Hence, cognitive effort could exist due to the field-of-view constraints inherent in virtual environments.