Construct and ecological validity of a new virtual reality measure of executive function; the Virtual Library Task

Abstract

Executive functions are commonly disrupted in people with a traumatic brain injury (TBI), with executive dysfunction having potentially devastating functional consequences. Neuropsychologists are increasingly being asked to make predictions regarding patients’ everyday functional abilities. Despite this, traditional neuropsychological measures of executive function lack sensitivity and ecological validity, whilst administration of newer functional measures such as the Multiple Errands Test can be time consuming, costly, difficult to standardize across settings, and may potentially violate patient privacy or create safety concerns for the patient, therapist, or general public. Virtual reality (VR) assessment paradigms that simulate real world environments and scenarios on computers have been suggested as a potential solution; however studies to date have lacked psychometric rigour. The overall aim of the present thesis was to develop a VR measure of executive function; the Virtual Library Task (VLT) and to compare the ecological and construct validity of this measure with existing neuropsychological measures of executive function. Thirty TBI participants and thirty control participants were administered the VLT, and a real life analogous task - the Real Library Task (RLT), and several neuropsychological measures of executive function: Verbal Fluency, Wisconsin Card Sorting Test – 64: Computer Version (WCST-64CV), the Brixton Spatial Anticipation Test and two more recent tests designed with ecological validity in mind; Zoo Map subtest and Modified Six Elements Test (MSET) from the Behavioural Assessment of the Dysexecutive Syndrome. Significant others for each participant also completed the Dysexecutive Questionnaire (DEX); a behavioural rating scale of everyday executive functioning. Results indicated that performances on the VLT and the RLT were significantly positively correlated. In addition the VLT measures (but none of the neuropsychological measures of executive function) were significant predictors of RLT scores. Despite the TBI group experiencing more everyday executive problems as reported by significant other DEX scores, only one of the five neuropsychological measures, the MSET, was able to differentiate the groups. The VLT successfully differentiated the groups, and was moderately correlated with neuropsychological measures of executive function and not correlated with a theoretically unrelated construct (immediate attention). Results indicated that only the VLT and the MSET were significantly correlated with DEX scores and both these measures significantly predicted everyday executive functioning. Overall the results across studies raise concerns regarding the use of the Verbal Fluency Test, WCST-64CV2, Brixton Spatial Anticipation Test and Zoo Map Test to accurately identify and predict executive function impairments in people with moderate to severe TBI. This is particularly concerning given that these measures are commonly used in clinical practice, with many clinicians interpreting adequate performance on them as patients’ having intact or unimpaired executive functions. In contrast the MSET appears to be both sensitive to executive dysfunction and capable of predicting everyday executive functioning. Results indicate that the VLT is also an ecologically valid measure of executive functioning and provide initial support for the construct validity of this test. The VLT has the advantage over the MSET by providing objective measurement of individual components of executive function. Further investigation of the VLT’s reliability and construct validity is recommended.