Perception Processing in Schizophrenia Patients and Normal Population

In the present study we investigate the relative contribution of component versus configural properties to face and object perception in schizophrenic and control subjects. To this end, we designed a series of experiments that are based on former work by Kimchi and Palmer (1982), and on my Ph.D. dissertation conducted under the supervision of Prof. Ruth Kimchi at the IIPDM.

Experiment 1 investigates the relative roles of local parts versus global configuration in object perception by applying a technique formalized by Kimchi and Palmer (1982). Participants are presented with hierarchical figures, in which larger figures (e.g., a "global" triangle) are constructed by smaller figures (e.g., "local" squares).

Participants (schizophrenics and controls) perform a similarity-judgment task. In each experimental trial, participants are presented with a triad of stimuli: a standard figure and two comparison figures. Participants are asked to determine which of the two comparison figures is more similar to the standard. Each figure consists of a global square or triangle made of local squares or triangles. In the "same configuration" comparison figure different elements are arranged in the same configuration as the standard figure. In the "same element" comparison figure, the same elements as in the standard figure are used, but are arranged in a different configuration. The number of elements used to construct the standard figure is varied. Previous data with normal perceivers (Kimchi \& Palmer, 1982) showed that as the number of elements increased (and their sizes decreased) perceivers tended to base their similarity judgments on the global configuration, rather than on its local elements. If schizophrenics are impaired in their ability to process configural information then their similarity judgments should mainly rely on the local elements rather than on the global configuration.

Experiment 2 examines the relative contribution of facial components (e.g., eyes, nose, mouth) versus configural properties (e.g., spacing between the eyes) to face perception, by applying Garner's speeded classification paradigm (Garner, 1974). We have previously applied this paradigm in my Ph.D. dissertation to study face perception in normal perceivers.

The stimulus set consists of four faces created by orthogonally combining components (eyes, nose, and mouth) and configural information (inter-eyes distance and nose-mouth distance). On each trial participants are asked to classify a face on either its components or its configural information, in two conditions. In the control condition only one facial dimension varies (e.g., components) while the other dimension (e.g., configural information) is held constant. In the filtering condition the two dimensions are varied orthogonally. Equal performance in the control and filtering conditions indicates that participants are able to selectively attend to the relevant dimension - the dimensions are processed independently. Poorer performance in the filtering condition than in the control condition - Garner interference - indicates that it is not possible to selectively attend to one dimension while ignoring another, irrelevant dimension - the dimensions are processed interactively.

Results with normal perceivers revealed a symmetric Garner interference with upright faces: participants were unable to selectively attend to components and ignore irrelevant variation in configural information, and vice versa. These results imply perceptual interdependence in processing components and configural information. If schizophrenics are impaired in their ability to process configural information then patients should be able to selectively attend to components and ignore irrelevant variation in configural information, but not vice versa. This pattern would indicate the relative dominance of components over configural properties in upright face processing of schizophrenia patients.