Abstract
Accommodation operates to optimise retinal-image contrast. There is a long-standing debate, however, about the role of different spatial frequencies. The fine-tuning hypothesis posits that high frequencies play an increasing role as the system approaches sharp focus. Yet empirical data show that adding mid-range frequencies (above 6–8 cpd) has little effect on accommodation performance. Recently, we presented an analysis of changes in retinal contrast with accommodation, based on real-eye optics, that reconciles these apparently contradictory ideas (MacKenzie, Hoffman, & Watt, 2010, JOV). We suggested that responses are driven by overall retinal-contrast changes with changes in accommodation, implying that the ‘contribution’ of particular frequencies depends on the relative strength of the signal they provide in a given stimulus. Here we test this model empirically. The contribution of different frequencies has typically been determined by measuring changes in variability of steady-state accommodation responses, or accommodation gain in response to modulations in focal distance. Interpreting such data is complicated because the optimal accommodative distance is unaffected by stimulus content. Instead, we used a multiple-focal-plane display to present different combinations of spatial frequencies at different focal distances. We used our model to make quantitative predictions about the biases in accommodation that would result. Accommodation responses were well predicted by the model. High spatial frequencies (~16 cpd) had a clear effect on accommodation when the signal at mid-range frequencies was weak (low contrast, narrow band). Consistent with previous experiments, however, higher frequencies had little effect when the mid-range signal was strong (high contrast, broadband). The results are consistent with the idea that accommodation is driven by overall changes in retinal contrast, summed across different frequency components. This suggests that the ‘role’ of different spatial frequencies in accommodation control is not fixed, but is an emergent property of the signal they provide in particular stimulus contexts.