Review of: Coy, B., O’Brien, W. H., Tabaczynski, T., Northern, J., & Carels, R. (2011). Associations between evaluation anxiety, cognitive interference and performance on working memory tasks. Applied Cognitive Psychology, 25(5), 823–832. doi:10.1002/acp.1765

Reviewed by Lauren Lee

There is a well-established “robust and reliable” body of research that links poorer performance on cognitive tasks with evaluation anxiety. Cognitive Interference Theory (CIT) proposed by Sarason and associates attempts to explain that association, using the term negative off-task self-dialogue to describe the interference caused by distraction (Sarason, 1984;Sarason, Sarason, & Pierce, 1990, 1995). This line of research is supportive of Beck’s cognitive view of anxiety as caused by a person’s increased likelihood of reacting to a threat, while underestimation their own problem solving abilities, paired with an over estimate of the threat posed. Many studies comparing levels of negative off-task dialogue with evaluation and off-task thoughts support CIT; however their used of correlational methods to compare the groups invites critique as establishing a link between variables without proving directionality of the relationship.

Working from Sarason’s theory, Eysenck and Calvo (1992) pursued the impact of negative off-task self-dialogue on working memory and subsequently their performance on cognitive tasks by priming the system used in assessing new information needed for cognitive tasks. Central Executive, an essential component of CIT, has 3 primary functions: inhibition, task switching, and encoding of incoming information. Additionally, it may serve as a relay to other auditory and visual processing systems. Coy et al. (2011) uses the verbal linguistic mentation model to explain the stronger association between verbal memory and negative off-task self-dialogue by interfering with the phonological loop of the working memory system. Due to this overload, the central executive processes will also be impacted downstream. Also proposed is that visual stimuli, because processed separately from verbal information, will be less effected by the negative off-task self-dialogue. Although Beilock et al., (2007) found data supportive of CIT, connecting evaluation anxiety to impaired phonological processing, central executive function, they were not able to establish to what extent the negative off-task dialogue or how much the stereotype threat presented, impacted performance.

while no experimental research has evaluated the link between evaluating anxiety and all three functions of working memory, the functions have been studied separately though also using correlational design. This study attempt to assess all three components of working memory, with experimental design, using assessments with establish reliability and validity. participants were ruled out based on potential for cognitive impairment due to organic cause, trauma, or substance. subsequently, 88 participant were compensated with credits towards a course requirement. Subjects were split into an anxiety provoking instruction group or a supportive instruction group and given the Digit Span task to assess the phonological loop impact, visuospatial memory task to assess the sketchpad component, and the Stroop Color-Word task to asses central executive function. Additionally, negative off-task self-dialogue was assessed pre- and post- assessment. Coy et al. (2011) are testing these hypotheses: “(a) participants who receive evaluation anxiety-inducing instructions will report significantly more negative off-task self-dialogue than participants who receive supportive instructions; (b) participants who receive evaluation anxiety-inducing instructions, relative to participants who receive supportive instructions, will demonstrate signifi- cantly decreased performance on the cognitive tasks that require the phonological loop and central executive components of working memory; (c) reports of negative off-task self-dialogue will be inversely correlated with performance on the phonological loop and central executive tasks and (d) negative off-task self-dialogue will mediate the relationship between evaluation anxiety and performance on the phonological loop and central executive tasks” (p. 824-825).


This team measured anxious activation through a heart-rate measure and test anxiety with the Revised Test Anxiety (RTA) Scale, though the questions were re-worded to assess state anxiety instead of trait anxiety (Benson, Moulin- Julian, Schwarzer, Seipp, & El-Zahhar, 1992). Negative off-task self-dialogue was measured using the Cognitive Interference Questionnaire (CIQ), which measures one’s negative off-task self-dialogue and the extent to which it interferes with concentration (Sarason & Stoops, 1978; Sarason, Sarason, Keefe, Hayes,& Shearin, 1986). Phonological loop function was assessed by Digit span forward and backward subtest and the visuospatial sketch-pad memory using visual memory span subtest, all from Wechsler’s Revised Memory Scale (WMS, Wechlser 1987). Central Executive function was assessed using the Stroop Color-Word reading task. Each assessment was chosen based on its established reliability and validity at measuring the proposed constructs. Evaluation anxiety induction instructions were developed by a blinded group of students who identified a group of questions presented as supportive or anxiety producing.


Student arrived at the Mindful Behavior Therapy and Psychophysiology Laboratory and filled out demographic questionnaires. Additionally, their baseline heart rate was established after a resting period and they were distributed randomly into wither the supportive instruction group or the anxiety provoking instruction group, then given the RTS, cognitive performance tests that were randomly ordered for each participant. Shortened versions of the instructions, both forms, were given prior to each subtest. Following the assessment, participants complete the CIQ and the RTA again.


There was no significant difference between demographics of the two groups. On anxiety measures, multivariate between group comparison established significantly elevated evaluation anxiety level with a large effect size (F (2, 85) = 23.48, p<.001). Heart rate reactivity scores were also significantly higher in the anxiety producing instruction group, suggesting that the instruction effectively elevated anxiety in the testing experience (F (5, 82) = 3.42, p =.007). Negative off-task self-dialogue was compared using one-way ANOVA and found significantly more negative off-task self-dialogue with the anxiety provoking instructions. The effect of evaluation anxiety on cognitive performance was established with a 2 X 5 MANOVA finding a significant main effect for the instruction variable (F (5,82) = 6.12, p<.001), with additional large, significant condition effects on specific subtests of Digit Span forward and backward and a slightly significant effect on the Stroop central executive exercise.

Negative off-task self-dialogue as a mediator

Structured equation modeling (SEM) was used to analyze the impact of negative off-task self-dialogue using the bootstrap estimation technique because of the small sample size. They found a positive and significantly large relationship between evaluation anxiety and the negative off-task self-dialogue with an inverse relationship with the phonological loop performance when compared to both evaluation anxiety and negative off-task self-dialogue. Controlling for negative off-task self-dialogue negated the significance between anxiety induction and phonological loop performance. An effect ratio of .410 points to 41 % mediation by negative off-task self-dialogue of variance between the anxiety induction condition and phonological loop performance. Effects on central executive performance were similar except for the non-significant relationship between negative off-task self-dialogue and central executive performance. Additionally, negative off-task self-dialogue was only a weak partial mediator between the anxiety induction group and central executive performance.



The authors’ predictions were supported. Specifically, there was a significant difference in outcome on each measure between the anxiety inducing instructions group and the supportive instructions group. These results are also consistent with previous research supporting relationship between negative off-task self-dialogue and evaluation anxiety and it impact of cognitive performance. Because of the random assignment of participants it can be postulated that the anxiety condition was induced by the instructions, triggering the negative off-task self-dialogue. Additionally, the hypothesis that the visuospatial sketchpad function would be less effected by the evaluation anxiety was supported. This supports Eysenck and associates argument that central executive and phonological loop functions are especially subject to the effects of anxiety and negative off-task self-dialogue. Because of the lack of valence of performance to the participant, it can be stated that interpretation of the test instructions can impact test performance generally, indicating a need for assessment of negative off-task self-dialogue in any testing environment to monitor test interpretation.

Their findings indicate a strong mediation effect of negative off-task self-dialogue on a person’s evaluation anxiety and phonological loop function, with less support for that mediation on the central executive tasks. These findings also support and add to the findings by Beilock et al. (2007) regarding stereotype threat in an assessment environment and reduced performance on verbal processing tasks caused by anxiety. Going beyond that, they studied the central executive and visual sketchpad functions of working memory as well as the mediation effect of negative off-task self-dialogue. Therefore, this research provides partial support for Cognitive Interference Theory (CIT) and pointed to probable mechanisms effecting cognitive performance. The authors found that, “elicitation of evaluation anxiety resulted in heightened negative off-task self-dialogue which, being a verbal-linguistic form of cognition, interfered most with performance on tasks that involve the processing of verbal stimuli in the working memory system” (p. 831).

Limitation and Future Directions

Because the CIQ was not administered after each cognitive task, it is not possible to establish difference in cognitive interference effect on the specific tasks or all the task combined. Also, demand characteristics of the participant might have influenced responses on the CIQ as they attempt to provide the desired responses, indicating a need for blind administrators and counter expectancy instructions. Due to the homogeneity of the sample, healthy college-aged students without cognitive impairment, the results have limited generalizability. The author suggests further research with clinical populations. Lastly, due to the limited mediation of negative off-task self-dialogue on central executive performance the authors wonders if the limited measure of task inhibition accessed by the Stroop Color-Word tasks is was resulted in the marginal significance. They propose further research with a broader range of executive tasks.