The Modality Shift Effect in Autism: Exploring Speed Accuracy Trade-Off and the Time Course of Crossmodal Switching.

Background: Previous research in neurotypicals (NTs) has shown that people can selectively attend to a sensory modality (vision, touch, hearing). This is reflected in switch costs during experimental tasks in which participants typically respond more slowly and less accurately to a target stimulus preceded by a stimulus in a different modality (crossmodal) relative to a target preceded by a stimulus in the same modality (ipsimodal). There are findings suggesting that shifts in attention between visual and auditory information are less effective in autism, reflected in increased switch costs compared with NTs (Courchesne et al., 1994; Williams, Goldstein, & Minshew, 2013). There are also contradictory findings suggesting audio-visual switch costs are comparable to NTs (Haigh et al., 2016; Murphy, Foxe, Peters, & Molholm, 2014).

Objectives: This study investigated automatic crossmodal switching between visual, tactile and auditory targets in autism for the first time. The time course of crossmodal switching was investigated; we expected that participants with autism would not show a reduction in switch costs with increased duration between the targets based on previous evidence of impaired disengagement of attention. We also sought to characterise any differences in speed- accuracy trade-offs during crossmodal switching using Drift Diffusion modelling (Ratcliff and Mc Koon, 2008).

Methods: Autistic adults (n =24) and NT controls (n = 24) matched for age, IQ, gender and handedness completed a speeded discrimination task (pulsed vs continuous) to visual, tactile and auditory targets. Targets were separated by 1,000, 1,250 or 2,000ms inter trial intervals (ITIs). For each target, reaction times were compared for ipsimodal and crossmodal trials across the three ITIs. The EZ diffusion model (Wagenmakers, van der Maas, & Grasman, 2007) was used to extract estimates of drift rate (quality of information extracted from the target), boundary separation (response conservativeness) and non-decision time (time taken to encode the target and prepare a response).

Results: Clear switch costs were observed in the reaction time data for each target modality, but did not differ between the groups. Against our expectations, switch costs were not reduced with increasing ITI for either group. For visual and auditory targets, switch costs were apparent in increased drift rates and reduced non-decision time. For visual targets, participants with ASC exhibited an increased boundary separation. Furthermore, participants with ASC required longer non-decision times when responding to tactile targets preceded by auditory.

Conclusions: Switch costs did not statistically differ between the groups and were not reduced with increased ITI. The importance of considering speed accuracy trade-offs in ASC was highlighted by differences in diffusion model parameters. In particular, the increased non-decision time when switching from auditory to tactile information suggests that the interaction between these sensory modalities may be affected in ASC and warrants further investigation.