Abstract
Purpose :
Serial effects, where the response to a particular stimulus is influenced by prior stimuli, have been reported in a range of psychophysical tasks. Yet, little is known regarding their impact on perimetry. We investigate serial dependence in response times (RTs) from frequency-of-seeing (FOS) data. We hypothesize that an increase in RT occurs for responses following unseen presentations.
Methods :
Analysis was performed on FOS data collected on 99 subjects with a range of visual field sensitivities (0 to 37 dB). Subjects took part in studies at Moorfields Eye Hospital (London, UK) and Dalhousie University (Halifax, Canada). The method of constant stimuli was used to present Goldmann size III stimuli (diameter, 0.43 degrees) for 200 ms. One eye per subject was tested, and between 80 and 400 stimuli were presented at 2 to 10 locations. The Psignifit toolbox (v.3.0, Frund et al. JOV 2011) was used to estimate the 50%-seen threshold for each location of the visual field. RTs were plotted against the difference between stimulus intensity and estimated threshold (△ contrast), and a robust (least absolute deviations) exponential regression model was built individually for each observer (Figure 1a). Residuals from the model were pooled and investigated as a function of responses to prior stimuli.
Results :
Thresholds were calculated from 460 FOS curves. The median predicted RT from the model was 540 ms (interquartile range [IQR], 485 – 600 ms) at threshold, and 450 ms (IQR, 410 – 525 ms) for stimuli 10 dB above threshold. Median RT differences (observed – predicted) were 30 ms for when none of the prior five stimuli were seen; 10 ms when 2 of 5 prior stimuli were seen; and -20 ms when all five prior stimuli were seen (p < 0.001; Figure 1b).
Conclusions :
Responses after several unseen stimuli have a substantially longer latency. This increase in RT may be the consequence of greater cognitive processing and/or a shift in response criteria. Sequence effects may need to be taken into account in the design of threshold algorithms for clinical visual field assessment.
This is an abstract that was submitted for the 2016 ARVO Annual Meeting, held in Seattle, Wash., May 1-5, 2016.