Pupillary responses on the visual backward masking task reflect general cognitive ability

Abstract

Cognitive processing efficiency requires both an ability to attend to task-relevant stimuli with quickness and accuracy, also while filtering distracting or task-irrelevant stimuli. This study investigated cognitive processing efficiency by using pupillary responses as an index of attentional allocation to relevant target and irrelevant masks on a visual backward masking task. The relationship between attentional allocation on this task and general cognitive ability on the scholastic aptitude test (SAT) was examined in college students (n=67). A principle components analysis of the pupillary response waveform isolated a late component that appeared to index the attentional demands associated with processing masks on the backward masking task. This pupillary response index of wasteful resource allocation to the mask accounted for significant variance in SAT scores over and above that accounted for by socio-economic status and target detection accuracy scores. Consistent with the neural efficiency hypothesis, individuals who allocated more resources to processing irrelevant information performed more poorly on cognitive ability tests.

Introduction

The neural efficiency hypothesis states that more intelligent individuals process information and solve problems more efficiently (i.e. with less mental effort) than less intelligent individuals (Davidson and Downing, 2000, Haier et al., 1992, Hendrickson, 1982a, Hendrickson, 1982b, Schafer, 1982). This hypothesis has received some support in the psychophysiological literature on pupillary responses. The extent of pupil dilation recorded during a cognitive task is a psychophysiological measure of task processing load and resource allocation, with larger pupil dilation reflecting greater processing load or mental effort (Beatty, 1982). Ahern and Beatty (1979) showed an association between pupillary responses and cognitive ability by showing that pupillary responses recorded in college students while they performed a multiplication task were negatively correlated with cognitive ability. That is, college students with lower scores on the scholastic aptitude test (SAT) exhibited greater pupil dilation to the multiplication problems than students with higher SAT scores. Consistent with the neural efficiency hypothesis, they concluded that individuals with greater cognitive ability process information with greater efficiency or less mental effort.

One information-processing task tapping speed and efficiency of processing is the visual backward masking task, which has received substantial notoriety in the intelligence literature (for reviews see Deary and Stough, 1996, Deary, 2000). The backward masking task is used to quantify the amount of time that information is passed through the sensory register. This task consists of a rapidly presented target stimulus (e.g. letters or different length lines), a varying length of vacant time (e.g. 20–700 ms) and a masking stimulus that typically completely covers the spatial presence of the target stimulus (Saccuzzo, 1993). Participants are typically asked to identify the target stimuli (e.g. which line is longer or what was the letter). Successful completion of the task, therefore, requires not only efficient processing of the target stimuli, but also the ability to filter the effects of the masking stimulus. Inspection time (IT), the amount of time needed for an individual to reliably perceive the target stimulus, has been touted as the best information-processing measure in terms of having a reliable, substantial correlation with performance on standard tests of psychometric intelligence (Deary and Stough, 1996). This measure derived from the backward masking paradigm accounts for approximately 20% of the variance in intelligence tests (Deary and Stough, 1996, Kranzler and Jensen, 1989, Longstreth et al., 1986, Nettlebeck et al., 1986).

In a previous study, we recorded pupil dilation responses in college students while they performed a visual backward masking task with 33, 50, 67, 117 and 317 ms stimulus onset asynchronies (SOA) between the target and mask stimuli and a no-mask condition (Verney et al., 2001). Pupil dilation was significantly greater during task performance (cognitive load) relative to a condition where participants passively viewed the stimuli (cognitive no-load), and there were no significant differences between SOA conditions during passive viewing of the stimuli (no-load). This finding further validates pupil dilation as an index of cognitive resource allocation. Moreover, significantly greater pupil dilation was found in the longest (317 ms) SOA condition compared to the no-mask condition. Dilation in all other SOA conditions did not exceed that of the no-mask condition. The only difference between the longest SOA condition and the no-mask condition was the presence of the mask. Therefore, this finding suggested that the presence of the mask increased task-processing load beyond that of target detection alone (no-mask condition) only in the longest (317) SOA condition. This finding was consistent with backward masking task models that suggest the mask demands extra processing resources, or a shifting and sharing of stimulus identification resources between the target and mask, only when the mask follows a target by more than approximately 120 ms (Loftus et al., 1988, Michaels and Turvey, 1979, Phillips, 1974).

The total pupil dilation response reflects the sum of all processing demands associated with the task. In an attempt to isolate the separate processing demands associated with specific task stimuli (e.g. targets and masks), a principle components analysis (PCA) was computed on the Verney et al. (2001) data set, as well as on pupillary response data sets from two additional backward masking task studies from our lab (Granholm and Verney, 2004, Verney, 2001). PCA is often used as a method of reducing the large number of data time points in psychophysiological data to a small number of meaningful factors. Three factors consistently emerged from the PCA analyses in all three of these studies, which appeared to isolate the specific resource demands associated with target and mask processing. The three factors formed a linear time course of the pupillary response waveform: (1) An early factor from approximately 0 to 0.7 s; (2) a middle factor from approximately 0.7 to 1.5 s; and (3) a late factor from approximately 1.5 to 3.0 s. The middle factor occurred in the time window when peak dilation responses to cognitive task stimuli are commonly found to reflect resource allocation to task performance (e.g. discriminating and evaluating the target lines; Beatty, 1982, Beatty and Lucero-Wagoner, 2000, Steinhauer and Hakerem, 1992). Middle factor dilation was smaller in conditions where target detection was poorest and larger in conditions where target detection was greatest. The middle factor, therefore, was interpreted as reflecting target processing. The late factor was interpreted as reflecting resources allocated to mask processing. In longer SOA conditions (i.e. >∼120 ms), when the masking stimulus becomes a distinct percept from the target stimulus (Michaels and Turvey, 1979, Phillips, 1974), late factor pupil dilation was significantly greater than in the no-mask condition. We interpreted this difference in the late factor between longer SOA masking conditions (with both target and mask) and the no-mask condition (containing only a target) as reflecting the additional processing demands of the mask. Therefore, the late factor dilation score could be used to measure resource allocation to mask processing.

The present study attempted to replicate and extend Ahern and Beatty's (1979) finding that cognitive task-evoked pupillary responses are negatively associated with general cognitive ability. In contrast to the Ahern and Beatty (1979) study, which used a higher-order processing task (multiplication), the visual backward masking task was used in this study to tap speed and efficiency of processing. This task was thought to be a better test of the neural efficiency hypothesis, because it has been used exclusively for this purpose in the intelligence literature and does not tap cognitive abilities directly measured on the SAT (e.g. math abilities). It was hypothesized that greater cognitive ability (SAT scores) would be associated with greater task detection accuracy. If confirmed, this would replicate previous findings that behavioral measures of information processing efficiency are strongly related to cognitive ability (Deary and Stough, 1996, Kranzler and Jensen, 1989). It was also hypothesized that pupillary dilation responses elicited by the task's non-informational masking stimulus (i.e. inefficient or wasteful mask processing) would significantly add to the prediction in SAT scores above that provided by detection accuracy and socio-economic status (SES). That is, consistent with Ahern and Beatty (1979), greater cognitive ability should be associated with less pupil dilation to the masking stimulus, especially in the longer SOA conditions where resource allocation to the mask is greatest. This finding would be consistent with the neural efficiency hypothesis that individuals with greater cognitive abilities perform tasks with less mental effort and do not wastefully allocate resources to task-irrelevant information (e.g. Cha and Merrill, 1994, Merrill and Taube, 1996, McCall, 1994).