Presentation is loading. Please wait.

Presentation is loading. Please wait.

An Eyetracking Analysis of the Effect of Prior Comparison on Analogical Mapping Catherine A. Clement, Eastern Kentucky University Carrie Harris, Tara Weatherholt,

Similar presentations


Presentation on theme: "An Eyetracking Analysis of the Effect of Prior Comparison on Analogical Mapping Catherine A. Clement, Eastern Kentucky University Carrie Harris, Tara Weatherholt,"— Presentation transcript:

1 An Eyetracking Analysis of the Effect of Prior Comparison on Analogical Mapping Catherine A. Clement, Eastern Kentucky University Carrie Harris, Tara Weatherholt, Barbara Burns University of Louisville Introduction An analogy is a cognitive tool that supports deep understanding of concepts, and the transfer of knowledge to new situations. In an analogy, two domains are compared that share a common theme or “relational structure”. Structural Alignment Identifying matching relational structure is described as "structural alignment": an interconnected system of relations among elements in a source domain is matched with a system of relations in a target (Gentner, 1989; Holyoak & Thagard, 1989). Obstacles to Alignment. Finding the matching structure can be difficult because: Similarities in relational structure are embedded in surface features that differ between domains. Corresponding objects are not similar. Existing contextual or object similarities may be distracting if they are unrelated to the shared structure. Prior Alignment May Facilitate Future Alignment. Structural alignment between analogs may increase attention to relational similarity, which may in turn facilitate future recognition of the analogous structure in new situations (e.g. Gick & Holyoak, 1989; Kotovsky & Gentner, 1996; Markman & Gentner, 1993). A Pilot Study of Eyetracking During Structural Alignment We explored the use of eyetracking to study the locus of attention during analogical mapping. Participants compared pairs of test scenes that shared a common structure. Corresponding objects in the structure did not match. However, the scenes also included irrelevant surface matches. Before mapping test scenes, “prior comparison” subjects compared scenes with the same structure as the test scenes. Eye-tracking analyses explored whether prior comparison subjects would be more likely than control subjects to attend to objects relevant to shared relational structure, and less likely to attend to irrelevant surface matches during analogical mapping. Materials Analogous Scene Pairs: Exposure and test pairs of analogous scenes based on two structures (A and B) are shown below. For Structure A one of the scenes being compared was the same at exposure and test. For Structure B: both scenes in the pair were new at test. Structure A Structure B Exposure Pair Exposure Pair Test Pair Test Pair Types of Object Matches: Structure Relevant Matches: objects played the same role in the shared structure, but were not similar at the surface level, e.g. the truck and the helicopter in Structure A. Surface Matches: objects had surface similarities but did not play a role in the shared structure, e.g. the dogs in Structure A. Only test pairs had surface matches as well as structure matches. Exposure pairs did not have surface matches. Procedure Exposure phase. Prior Comparison Subjects (n=11) were asked to view exposure pairs one at a time and to rate their similarity on a scale of 1-5 (non- analogous pairs were also shown). Control Subjects (n=12) were told to study individual scenes in a pair for 15 seconds for a later memory test. Analogous scenes were not shown contiguously. Test phase. All subjects rated the similarity of the test pairs.. Eye movement recording: eye movements were monitored with the ASL Model 504 eye tracking system (Applied Sciences Laboratory). Stimuli were displayed on a 16” monitor and eye movements were recorded using Gaze Tracker software (Eye Response Technologies). Fixations consisted of a minimum of 3 gazepoints; the minimum time for a fixation was.20 second. Sample Output

2 Results Fixations to Structure-Relevant and Surface-Similar Objects in Each Group Proportion of fixations (Figure 1). For both structures, both groups looked more often at structure- relevant objects. For Structure A, control subjects looked somewhat more at surface similar objects than did prior comparison subjects (t(21) = -2.66, p=.104). For Structure B, the groups were the same in fixations to surface-similar objects overall. However, control subjects looked significantly more at a particular kind of surface-similar object: those objects that were shown in isolation, and were not part of any action in the scenes (the cats and the single flowers), t(21) = -2.45, p=.02. Number of each type of object ever observed. We counted the number of structure-relevant and surface-similar objects in each scene that a subject fixated on at least once (Figure 2). Again control subjects looked at more surface similar objects for Structure A, t(21)= -2.5, p=.02 and at more isolated similar objects in Structure B t(21)= - 1.77, p =.09. Movement Patterns. Analyses considered subjects’ patterns of eye movements while comparing the scenes in a pair. No group differences were found for these analyses. First, we classified successive fixations in terms of whether they occurred between objects within a scene, within a single object, or between scenes in a pair. As shown in Figure 3, overall, more movements occurred within a scene than between scenes, F(1,21) = 9.077, p =.007 (Structure A); F(1,21) = 12.39, p =.002 (Structure B). We also looked specifically at whether the successive fixations between scenes occurred between corresponding structures, or between surface-similar objects (or between other objects). As shown in Figure 4, a relatively small proportion of movements occurred between either type of match, but overall significantly more movements occurred between corresponding structures than between surface matches, F(1,21) = 21.76, p =.000 for Structure A and F(1,21) = 4.6, p =.04 Structure B. Exposure and Test Phase Comparisons One scene was viewed during both the exposure and test phase by all subjects (the farm scene for Structure A). Fixations to this scene are shown in Figure 5. These data show: Even outside the context of a comparison, i.e. when control subjects saw the scene alone during exposure, the relative proportion of fixations was greater to structure-relevant objects than to those objects that would later have surface matches with the test analog. The presence of the test analog that included surface matches to the farm scene did not significantly increase fixations to the matching objects. This is shown by comparing fixations by prior comparison subjects on the "surface similar " objects during exposure, when surface matches were not present, to test when the matches were present. Summary and Discussion This pilot study used eyetracking to examine how prior experience matching analogous scenes would affect attention during analogical mapping. Eyetracking was monitored as subjects rated the similarity of analogous scenes. 1. With and without prior comparison, subjects fixated more on objects related to the shared relational structure of analogous scenes, than on objects that shared only surface features. Prior comparison had a slight tendency to further reduce attention to surface similarity. It should be noted that prior comparison did not increase the similarity ratings given to the analogous scenes. 2. The tendency for prior-comparison subjects to specifically look less at isolated object matches (for Structure B) suggests that these subjects may have developed a set to ignore information in a scene that is unlikely to have relational similarity. 3. Movement patterns suggest that more attention was devoted to comprehending an individual scene than to making object matches between scenes. More movements occurred within scenes than between scenes, and a relatively small proportion of movements occurred directly between either corresponding structures or similar objects. 4. A difficulty with using eyetracking to study analogy is that although we can illustrate that subjects fixate more on relevant objects, decisions about relational similarity may not be reflected in eye movements. However, different materials or tasks may more easily reveal attention to relational similarity, or more easily track the decision- making process. 5. Future research might examine the effect of prior comparison on stimulus representation. That is, eye movements might be used to examine whether attention to a scene, outside of the context of a comparison task, has been altered by previously having compared that scene to an analog. Controlling for the number of each type of object. Both groups fixated significantly more on structure-relevant than surface-similar objects even when the number of objects in the scene of each type is taken into account (F (1, 21) = 7.6, p =.012, for Structure A; (F (1, 21) = 23.7, p =.000, for Structure B. Correlations with Similarity Ratings The frequency of fixations to structure-relevant and surface-similar objects was not correlated with similarity ratings of test scenes. Also groups did not differ in similarity ratings of test scenes.


Download ppt "An Eyetracking Analysis of the Effect of Prior Comparison on Analogical Mapping Catherine A. Clement, Eastern Kentucky University Carrie Harris, Tara Weatherholt,"

Similar presentations


Ads by Google