Despite this caveat, the study provides an important challenge to our understanding of the role of
gain fields in spatial representation and computation. A number of outstanding questions remain. First, are these findings robust across different cortical areas known to contain eye-position signals, or are they specific to LIP? Another recent study of gain field dynamics (Morris et al., 2012) shows similar lags for eye-position signals in LIP, such that most LIP neurons do not provide reliable information about eye position until around 200 ms after an eye movement. Interestingly, while this result is consistent with Xu et al. (2012), these results were not reproduced in nearby dorsal visual areas VIP, MT, and MST. Instead, eye-position signals in these areas appear to learn more update much more rapidly, right around the time of the saccade and in some cases even slightly before the movement begins. These apparent inconsistencies in the temporal dynamics of gain fields across cortical areas produce a tension that requires resolution. Nevertheless, caution must be exercised in PD-1/PD-L1 inhibitor drugs drawing too strong a conclusion, since the paradigms differ in substantial ways: Morris et al. (2012) investigate eye-position modulation during static
fixation, whereas Xu et al. (2012) examine modulation in response to a visual target. A second outstanding question is whether the findings about the dynamics of eye-position gain fields in LIP apply to other motor systems or are specific to the oculomotor system. The authors imply that their findings have wide application, but this remains to be seen. Unique features of the oculomotor system could weigh against the extensibility of Xu et al.’s reported results. Most prominently, the oculomotor system—unlike many other motor systems—does not generally require an explicit computation of target during location in supraretinal (e.g., head-centered) coordinates, since typically only the retinal difference vector (the difference between the fovea and the retinal position of the target) is required for saccade programming. Consequently, the use or disuse of eye-position gain fields
for computations related to saccade programming might not accurately reflect how other motor systems use them, especially where reference frame transformations are required (Pouget and Snyder, 2000). Finally, Xu et al.’s results should lead researchers in the field to reflect more broadly about what other roles (if any) gain fields might play in motor planning and sensorimotor transformations. Given their widespread presence throughout the brain, it is incumbent upon the field to embrace the purely negative answer that they play no functional role only as a last resort. Xu et al. (2012) hypothesize that the temporal properties of these eye-position signals, while unsuited for use in real-time saccade programming, might be deployed in a more ancillary way as a kind of feedback to calibrate motor efference copy signals.