A model for the distortions of space and time perception during saccades
Perception of space and time undergoes strong distortions at the time of saccadic eye movements. Spatial distances are compressed towards the saccadic target while temporal intervals are underestimated [1-3]. We simulated these effects in a two-stage model where visual stimuli are encoded within a spatiotopic map and decoded by matching activation patterns with a set of templates learned in steady-fixation conditions. The retinotopic input to brief stimuli is given by an impulse response function that peaks after 25 ms and decays to half peak value in 65 ms, extending over the duration of the saccade. The input propagates within the spatiotopic layer with a dissipative automata whose maximum spreading velocity of about 1°/ms. At the encoding stage, recalibration of the spatiotopicity for a shift of gaze is obtained by correcting the retinotopic input with an internal signal representing eye position (corollary discharge), which starts 60 ms before the saccade and is complete at saccadic offset. The recalibration biases the automata spreading function parallel to the direction of the saccade, without altering the decoding stage. A given external stimulus elicits the same response in the map when presented long before or long after a saccade. In the proximity of a saccade, when remapping and/or retinal slip occur, activation patterns are unusual and smeared in both space and time. The second stage of the system decodes the activation of the spatiotopic map, determining the location and timing of the stimulus as well as the relative spatial and temporal distance of a couple of stimuli. The system performs veridically both for space and time localization in fixation conditions. However, at the time of saccades, the classifier is faced with unusual activation patterns causing it to distort both spatial and temporal metrics. The best match for a couple of peri-saccadic stimuli results in a compression of distances of about 50% both in space and time, consistently with psychophysical data. Absolute perceived time is also altered, so perisaccadic targets are delayed by about 100 ms; time momentarily stops or even inverts at the onset of remapping, again simulating well the perceptual data. The model is physiologically plausible given that remapping signals are very common in associative and parietal cortex and spatiotopic cortical maps have been described in VIP and area V6 of the monkey [4-6]. The model demonstrates that localization is achieved by a simultaneous optimal decoding of space and time. When this strategy is applied at the time of saccadic recalibration, the price is a transient distortion of the metrics for perisaccadic stimuli.
1. J. Ross, M.C. Morrone, and D.C. Burr. Nature, 384: 598 (1997)
2. M.C. Morrone, J. Ross and D.C. Burr. Nat Neurosci, 8: 950 (2005)
3. P. Binda, D.C. Burr and M.C. Morrone. Perception 36S:112 (2007)
4. JR. Duhamel, C.L. Colby, M.E. Goldberg. Science. 255:90 (1992)
5. JR. Duhamel et al. Nature 389:845 (1997)
6. C. Galletti, P. Battaglini, P. Fattori. Exp. Brain Res. 96:221 (1993) Funded by EU 6th & 7th Framework Programme MEMORY and STANIB
Computational and systems
neuroscience 2009, Salt Lake City, UT, USA, 26 Feb - 3 Mar, 2009.
(2009). A model for the distortions of space and time perception during saccades.
Front. Syst. Neurosci.
Computational and systems
10 Feb 2009;
10 Feb 2009.
Guido Marco Cicchini, firstname.lastname@example.org