A sample of 519 subjects were randomly drawn from a population of 130,000 school children of grades 3 to 6 in Taipei City. A method similar to that used by Haber and Haber (1964) was adopted to screen the eidetic children from this sample. Results indicated that if the ”Strict Criteria,” the same criteria used by the Habers, were used, 3.3% of the 519 subjects were identified as eidetic children. By the ”Less Strict Criteria,” the percentage was 5.8%. The former percentage (3.3%) was significantly smaller than that found in the Habers´ study (8%). Both these studies, however, revealed that eidetic imagery is not a widely prevalent phenomenon.

Travers´ assumption was tested in four experiments. The rationale upon which these experiments were based was that because of a retarded erasure mechanism, images from antecedent stimuli continue to persist for a relatively long time and accumulate, overlap, or superimpose with the images, or even the percepts, of succeeding stimuli. Thirty eidetic children selected by the ”Less Strict Criteria” and 30 non-eidetic children were used as subjects. 

Results of these experiments can be stated as follows: (1) The mean reaction time to the offset of the visual stimulus for the eidetic children was significantly longer than that for the non-eidetic children. (2) In a task requiring the subjects to search out the absence of target stimuli, the mean search time was not significantly different for the groups established by the ”Less Strict Criteria,” but eidetic subjects selected by the ”Strict Criteria” required significantly more time in this task than their counterparts. (3) The eidetic children demonstrated a persisting visual image in several ways. They reported a compound picture when two slide pictures were presented in sequence. They correctly identified figures that were ”hidden” in sequentially presented dot-patterns. Also, they reported a stereoscopic effect when the left and right-eye views were presented to the left and right eyes without the advantage of a stereoscope. Further, they reported the perception of an expanding spiral after viewing a rotating-contracting spiral while the stationary spiral had in fact been removed. (4) In EEG tests, the post-stimulation recovery time of the eidetic children was significantly longer than that for the non-eidetic children. The post-stimulation alpha index for the eidetic children was significantly smaller than that for the nor-eidetic children.

While deductions based upon Travers´ assumption were generally confirmed, there were several results that indicated that the construct of a deficient erasure mechanism was not sufficient to account for the phenomena associated with eidetic imagery. The limitations of Travers´ assumption were discussed and a new model for the phenomena of eidetic imagery tentatively proposed.