Domain-general object encoding ability explains shared variance between perception and memory factors

Domain-general object recognition ability (o) is an individual’s capacity to tell apartsimilar objects. O has been measured using tasks that make strong demands on eitherperception or memory. These types of test correlate with one another, suggesting that theyboth require precise encoding of high-level visual details. However, the assumption that tests ofeach type differentially measure perception and memory has not been tested, nor has thetheoretically critical assumption that tests of each type tap into the same high-level visual abilitywhen other relevant abilities are controlled. A large online sample completed a battery ofperceptual and memory type object recognition tests alongside tests of low-level perception,memory, and general intelligence. We specified preregistered structural equation models.Results indicated that perception and memory o tests form two separable factors, each tappinginto concurrently measured perceptual and memory abilities respectively. However, also in linewith our predictions, these two factors were very strongly correlated, and remained so whengeneral intelligence, and perceptual and memory abilities were regressed out. We argue thatthe remaining covariance between the two factors is attributable to o. While our results suggestthat general intelligence explains a very substantial share of the ability to make high-level visualdiscriminations, they also suggest that it does not explain all of this ability. The results also haveimplications for interpretation of prior research that distinguishes face perception and facememory.