A generative learning model for saccade adaptation.
ABSTRACT: Plasticity in the oculomotor system ensures that saccadic eye movements reliably meet their visual goals-to bring regions of interest into foveal, high-acuity vision. Here, we present a comprehensive description of sensorimotor learning in saccades. We induced continuous adaptation of saccade amplitudes using a double-step paradigm, in which participants saccade to a peripheral target stimulus, which then undergoes a surreptitious, intra-saccadic shift (ISS) as the eyes are in flight. In our experiments, the ISS followed a systematic variation, increasing or decreasing from one saccade to the next as a sinusoidal function of the trial number. Over a large range of frequencies, we confirm that adaptation gain shows (1) a periodic response, reflecting the frequency of the ISS with a delay of a number of trials, and (2) a simultaneous drift towards lower saccade gains. We then show that state-space-based linear time-invariant systems (LTIS) represent suitable generative models for this evolution of saccade gain over time. This state-equation algorithm computes the prediction of an internal (or hidden state-) variable by learning from recent feedback errors, and it can be compared to experimentally observed adaptation gain. The algorithm also includes a forgetting rate that quantifies per-trial leaks in the adaptation gain, as well as a systematic, non-error-based bias. Finally, we study how the parameters of the generative models depend on features of the ISS. Driven by a sinusoidal disturbance, the state-equation admits an exact analytical solution that expresses the parameters of the phenomenological description as functions of those of the generative model. Together with statistical model selection criteria, we use these correspondences to characterize and refine the structure of compatible state-equation models. We discuss the relation of these findings to established results and suggest that they may guide further design of experimental research across domains of sensorimotor adaptation.
Project description:Repeated exposure to a consistent trans-saccadic step in the position of the saccadic target reliably produces a change of saccadic gain, a well-established trans-saccadic motor learning phenomenon known as saccadic adaptation. Trans-saccadic changes can also produce perceptual effects. Specifically, a systematic increase or decrease in the size of the object that is being foveated changes the perceptually equivalent size between fovea and periphery. Previous studies have shown that this recalibration of perceived size can be established within a few dozen trials, persists overnight, and generalizes across hemifields. In the current study, we use a novel adjustment paradigm to characterize both temporally and spatially the learning process that subtends this form of recalibration, and directly compare its properties to those of saccadic adaptation. We observed that sinusoidal oscillations in the amplitude of the trans-saccadic change produce sinusoidal oscillations in the reported peripheral size, with a lag of under 10 trials. This is qualitatively similar to what has been observed in the case of saccadic adaptation. We also tested whether learning is generalized to the mirror location on the opposite hemifield for both size recalibration and saccade adaptation. Here the results were markedly different, showing almost complete generalization for recalibration and no generalization for saccadic adaptation. We conclude that perceptual and visuomotor consequences of trans-saccadic changes rely on learning mechanisms that are distinct but develop on similar time scales.
Project description:Control of saccadic gain is often viewed as a simple compensatory process in which gain is adjusted over many trials by the postsaccadic retinal error, thereby maintaining saccadic accuracy. Here, we propose that gain might also be changed by a reinforcement process not requiring a visual error. To test this hypothesis, we used experimental paradigms in which retinal error was removed by extinguishing the target at the start of each saccade and either an auditory tone or the vision of the target on the fovea was provided as reinforcement after those saccades that met an amplitude criterion. These reinforcement procedures caused a progressive change in saccade amplitude in nearly all subjects, although the rate of adaptation differed greatly among subjects. When we reversed the contingencies and reinforced those saccades landing closer to the original target location, saccade gain changed back toward normal gain in most subjects. When subjects had saccades adapted first by reinforcement and a week later by conventional intrasaccadic step adaptation, both paradigms yielded similar degrees of gain changes and similar transfer to new amplitudes and to new starting positions of the target step as well as comparable rates of recovery. We interpret these changes in saccadic gain in the absence of postsaccadic retinal error as showing that saccade adaptation is not controlled by a single error signal. More generally, our findings suggest that normal saccade adaptation might involve general learning mechanisms rather than only specialized mechanisms for motor calibration.
Project description:Saccadic adaptation is the motor learning process that keeps saccade amplitudes on target. This process is eye position specific: amplitude adaptation that is induced for a saccade at one particular location in the visual field transfers incompletely to saccades at other locations. In our current study, we investigated wether this eye position signal corresponds to the initial or to the final eye position of the saccade. Each case would have different implications on the mechanisms of adaptation. The initial eye position is not directly available, when the adaptation driving post saccadic error signal is received. On the other hand the final eye position signal is not available, when the motor command for the saccade is calculated. In six human subjects we adapted a saccade of 15 degree amplitude that started at a constant position. We then measured the transfer of adaptation to test saccades of 10 and 20 degree amplitude. In each case we compared test saccades that matched the start position of the adapted saccade to those that matched the target of the adapted saccade. We found significantly more transfer of adaptation to test saccades with the same start position than to test saccades with the same target position. The results indicate that saccadic adaptation is specific to the initial eye position. This is consistent with a previously proposed effect of gain field modulated input from areas like the frontal eye field, the lateral intraparietal area and the superior colliculus into the cerebellar adaptation circuitry.
Project description:The improvement of motor behavior, based on experience, is a form of learning that is critically dependent on the cerebellum. A well studied example of cerebellar motor learning is short-term saccadic adaptation (STSA). In STSA, information on saccadic errors is used to improve future saccades. The information optimizing saccade metrics is conveyed by Purkinje cells simple spikes (PC-SS) because they are the critical input to the premotor circuits for saccades. We recorded PC-SS of monkeys undergoing STSA to reveal the code used for improving behavior. We found that the discharge of individual PC-SS was unable to account for the behavioral changes. The PC-SS population burst (PB), however, exhibited changes that closely paralleled the qualitatively different changes of saccade kinematics associated with gain-increase and gain-decrease STSA, respectively. Gain-increase STSA, characterized by an increase in saccade duration, replicates the relationship between saccade duration and the end of the PB valid for unadapted saccades. In contrast, gain-decrease STSA, which sports normal saccade duration but reduced saccadic velocity, is characterized by a PB that ends well before the adapted saccade. This suggests that the duration of normal as well as gain-increased saccades is determined by appropriately setting the end of PB end. However, the duration of gain-decreased saccades is apparently not modified by the cerebellum because the PB signals ends too early to determine saccade end. In summary, STSA, and most probably cerebellar-dependent learning in general, is based on optimizing the shape of a PC-SS population response.
Project description:To perform a saccadic response to a visual stimulus, a 'sensorimotor transformation' is required (i.e., transforming stimulus location into a motor command). Where in the brain is this accomplished? While previous monkey neurophysiology and human fMRI studies examined either parietal cortex or frontal eye field, we studied both of these regions simultaneously using magnetoencephalography (MEG). Nineteen healthy participants performed a pseudorandom series of prosaccades and antisaccades during MEG. Antisaccades require a saccade in the direction opposite a suddenly appearing stimulus. We exploited this dissociation between stimulus and saccadic direction to identify cortical regions that show early activity for a contralateral stimulus and late activity for a contralateral saccade. We found that in the left hemisphere both the intraparietal sulcus and the frontal eye field showed a pattern of activity consistent with sensorimotor transformation - a transition from activity reflecting the direction of the stimulus to that representing the saccadic goal. These findings suggest that sensorimotor transformation is the product of coordinated activity across the intraparietal sulcus and frontal eye field, key components of a cortical network for saccadic generation.
Project description:Sensorimotor learning adapts motor output to maintain movement accuracy. For saccadic eye movements, learning also alters space perception, suggesting a dissociation between the performed saccade and its internal representation derived from corollary discharge (CD). This is critical since learning is commonly believed to be driven by CD-based visual prediction error. We estimate the internal saccade representation through pre- and trans-saccadic target localization, showing that it decouples from the actual saccade during learning. We present a model that explains motor and perceptual changes by collective plasticity of spatial target percept, motor command, and a forward dynamics model that transforms CD from motor into visuospatial coordinates. We show that learning does not follow visual prediction error but instead a postdictive update of space after saccade landing. We conclude that trans-saccadic space perception guides motor learning via CD-based postdiction of motor error under the assumption of a stable world.
Project description:Whenever we move our eyes, some visual information obtained before a saccade is combined with the visual information obtained after a saccade. Interestingly, saccades rarely land exactly on the saccade target, which may pose a problem for transsaccadic perception as it could affect the quality of postsaccadic input. Recently, however, we showed that transsaccadic feature integration is actually unaffected by deviations of saccade landing points. Possibly, transsaccadic integration remains unaffected because the presaccadic shift of attention follows the intended saccade target and not the actual saccade landing point during regular saccades. Here, we investigated whether saccade landing point errors can in fact alter transsaccadic perception when the presaccadic shift of attention follows the saccade landing point deviation. Given that saccadic adaptation not only changes the saccade vector, but also the presaccadic shift of attention, we combined a feature report paradigm with saccadic adaptation. Observers reported the color of the saccade target, which occasionally changed slightly during a saccade to the target. This task was performed before and after saccadic adaptation. The results showed that, after adaptation, presaccadic color information became less precise and transsaccadic perception had a stronger reliance on the postsaccadic color estimate. Therefore, although previous studies have shown that transsaccadic perception is generally unaffected by saccade landing point deviations, our results reveal that this cannot be considered a general property of the visual system. When presaccadic shifts of attention follow altered saccade landing points, transsaccadic perception is affected, suggesting that transsaccadic feature perception might be dependent on visual spatial attention.
Project description:Eye movements create an ever-changing image of the world on the retina. In particular, frequent saccades call for a compensatory mechanism to transform the changing visual information into a stable percept. To this end, the brain presumably uses internal copies of motor commands. Electrophysiological recordings of visual neurons in the primate lateral intraparietal cortex, the frontal eye fields, and the superior colliculus suggest that the receptive fields (RFs) of special neurons shift towards their post-saccadic positions before the onset of a saccade. However, the perceptual consequences of these shifts remain controversial. We wanted to test in humans whether a remapping of motion adaptation occurs in visual perception.The motion aftereffect (MAE) occurs after viewing of a moving stimulus as an apparent movement to the opposite direction. We designed a saccade paradigm suitable for revealing pre-saccadic remapping of the MAE. Indeed, a transfer of motion adaptation from pre-saccadic to post-saccadic position could be observed when subjects prepared saccades. In the remapping condition, the strength of the MAE was comparable to the effect measured in a control condition (33±7% vs. 27±4%). Contrary, after a saccade or without saccade planning, the MAE was weak or absent when adaptation and test stimulus were located at different retinal locations, i.e. the effect was clearly retinotopic. Regarding visual cognition, our study reveals for the first time predictive remapping of the MAE but no spatiotopic transfer across saccades. Since the cortical sites involved in motion adaptation in primates are most likely the primary visual cortex and the middle temporal area (MT/V5) corresponding to human MT, our results suggest that pre-saccadic remapping extends to these areas, which have been associated with strict retinotopy and therefore with classical RF organization. The pre-saccadic transfer of visual features demonstrated here may be a crucial determinant for a stable percept despite saccades.
Project description:The cerebellar vermis (lobules VI-VII) has been implicated in both postmortem and neuroimaging studies of autism spectrum disorders (ASD). This region maintains the consistent accuracy of saccadic eye movements and plays an especially important role in correcting systematic errors in saccade amplitudes such as those induced by adaptation paradigms. Saccade adaptation paradigms have not yet been used to study ASD. Fifty-six individuals with ASD and 53 age-matched healthy controls performed an intrasaccadic target displacement task known to elicit saccadic adaptation reflected in an amplitude reduction. The rate of amplitude reduction and the variability of saccade amplitude across 180 adaptation trials were examined. Individuals with ASD adapted slower than healthy controls, and demonstrated more variability of their saccade amplitudes across trials prior to, during and after adaptation. Thirty percent of individuals with ASD did not significantly adapt, whereas only 6% of healthy controls failed to adapt. Adaptation rate and amplitude variability impairments were related to performance on a traditional neuropsychological test of manual motor control. The profile of impaired adaptation and reduced consistency of saccade accuracy indicates reduced neural plasticity within learning circuits of the oculomotor vermis that impedes the fine-tuning of motor behavior in ASD. These data provide functional evidence of abnormality in the cerebellar vermis that converges with previous reports of cellular and gross anatomic dysmorphology of this brain region in ASD.
Project description:Saccadic eye movements provide a valuable model to study the brain mechanisms underlying motor learning. If a target is displaced surreptitiously while a saccade is underway, the saccade appears to be in error. If the error persists gradual neuronal adjustments cause the eye movement again to land near the target. This saccade adaptation typically follows an exponential time course, i.e., adaptation speed slows as adaptation progresses, indicating that the sensitivity to error decreases during adaptation. Previous studies suggested that the superior colliculus (SC) sends error signals to drive saccade adaptation. The objective of this study is to test whether the SC error signal is related to the decrease in the error sensitivity during adaptation. We show here that the visual activity of SC neurons, which is induced by a constant visual error that drives adaptation, decreases during saccade adaptation. This decrease of sensitivity to visual error was not correlated with the changes of primary saccade amplitude. Therefore, a possible interpretation of this result is that the reduction of visual sensitivity of SC neurons contributes an error sensitivity signal that could help control the saccade adaptation process.