J Satel

Modeling inhibition of return as short-term depression of early sensory input to the superior colliculus.

Inhibition of return (IOR) is an orienting phenomenon characterized by slower behavioral responses to spatially cued, relative to uncued targets, when the cue-target onset asynchronies (CTOAs) are long enough that cue-elicited attentional capture has dispersed. Here, we implement a short-term depression (STD) account of IOR within a neuroscientifically based dynamic neural field model (DNF) of the superior colliculus (SC). In addition to the prototypical findings in the cue-target paradigm (i.e., the biphasic pattern of behavioral enhancement at short CTOAs and behavioral costs at long CTOAs), a variety of findings in the literature are generated with this model, including IOR in averaging saccades and the co-existence of IOR and endogenous orienting at the same location. Many findings that cannot be accommodated by this model could be accounted for by incorporating cortical contributions.

Oculomotor inhibition of return: How soon is it “recoded” into spatiotopic coordinates?

When, in relation to the execution of an eye movement, does the recoding of visual information from retinotopic to spatiotopic coordinates happen? Two laboratories seeking to answer this question using oculomotor inhibition of return (IOR) have generated different answers: Mathôt and Theeuwes (Psychological Science 21:1793–1798, 2010) found evidence for the initial coding of IOR to be retinotopic, while Pertzov, Zohary, and Avidan (Journal of Neuroscience 30:8882–8887, 2010) found evidence for spatiotopic IOR at even shorter postsaccadic intervals than were tested by Mathôt and Theeuwes (Psychological Science 21:1793–1798, 2010). To resolve this discrepancy, we conducted two experiments that combined the methods of the previous two studies while testing as early as possible. We found early spatiotopic IOR in both experiments, suggesting that visual events, including prior fixations, are typically coded into an abstract, allocentric representation of space either before or during eye movements. This type of coding enables IOR to encourage orienting toward novelty and, consequently, to perform the role of a foraging facilitator.

Examining the dissociation of retinotopic and spatiotopic inhibition of return with event-related potentials.

Inhibition of return (IOR) is thought to reflect a mechanism that biases orienting which, under some circumstances, reduces perceptual processing at previously processed locations. Studies using event-related potentials (ERPs) have generally revealed that IOR is accompanied by an amplitude reduction of early sensory ERP components (e.g., P1). While behavioral studies suggest that IOR may be represented in both spatiotopic and retinotopic coordinates, all previous ERP studies have used the prototypical spatial cueing paradigm and have thus confounded retinotopic and spatiotopic reference frames. Because of this confound it is unknown whether the P1 reduction that has been associated with IOR will be observed in retinotopic or spatiotopic coordinates when these are dissociated. The current experiment investigated whether the P1 component would be modulated by IOR when the retinotopic and spatiotopic reference frames were dissociated by an eye movement between cue and target onset. Strong spatiotopic IOR was found to be accompanied by a negative difference (Nd) in the 200-300 ms time window, while a P1 reduction was absent, suggesting that P1 reductions do not provide an accurate reflection of IOR.

The effects of ignored versus foveated cues upon inhibition of return: An event-related potential study

Taylor and Klein (Journal of Experimental Psychology: Human Perception and Performance 26:1639–1656, 2000) discovered two mutually exclusive “flavors” of inhibition of return (IOR): When the oculomotor system is “actively suppressed,” IOR affects input processes (the perception/attention flavor), whereas when the oculomotor system is “engaged,” IOR affects output processes (the motor flavor). Studies of brain activity with ignored cues have typically reported that IOR reduces an early sensory event-related potential (ERP) component (i.e., the P1 component) of the brain’s response to the target. Since eye movements were discouraged in these experiments, the P1 reduction might be a reflection of the perception/attention flavor of IOR. If, instead of ignoring the cue, participants made a prosaccade to the cue (and then returned to fixation) before responding to the target, the motor flavor of IOR should then be generated. We compared these two conditions while monitoring eye position and recording ERPs to the targets. If the P1 modulation is related to the perceptual/attentional flavor of IOR, we hypothesized that it might be absent when the motoric flavor of IOR was generated by a prosaccade to the cue. Our results demonstrated that target-related P1 reductions and behavioral IOR were similar, and significant, in both conditions. However, P1 modulations were significantly correlated with behavioral IOR only when the oculomotor system was actively suppressed, suggesting that P1 modulations may only affect behaviorally exhibited IOR when the attentional/perceptual flavor of IOR is recruited.

Sensory and motor mechanisms of oculomotor inhibition of return.

We propose two explicit mechanisms contributing to oculomotor inhibition of return (IOR): sensory and motor. Sensory mechanism: repeated visual stimulation results in a reduction in visual input to the superior colliculus (SC); consequently, saccades to targets that appear at previously stimulated retinotopic locations will have longer latencies than those that appear at unstimulated locations. Motor mechanism: the execution of a saccade results in asymmetric activation in the SC; as a result, saccades that reverse vectors will have longer latencies than those that repeat vectors. In the IOR literature, these two mechanisms correspond to IOR effects observed following covert exogenous orienting and overt endogenous orienting, respectively. We predict that these two independent mechanisms will have additive effects, a prediction that is confirmed in a behavioral experiment. We then discuss how our theory and findings relate to the oculomotor IOR literature.

On the role of eye movement monitoring and discouragement on inhibition of return in a go/no-go task.

Inhibition of return (IOR) most often describes the finding of increased response times to cued as compared to uncued targets in the standard covert orienting paradigm. A perennial question in the IOR literature centers on whether the effect of IOR is on motoric/decision-making processes (output-based IOR), attentional/perceptual processes (input-based IOR), or both. Recent data converge on the idea that IOR is an output-based effect when eye movements are required or permitted whereas IOR is an input-based effect when eye movements are monitored and actively discouraged. The notion that the effects of IOR may be fundamentally different depending on the activation state of the oculomotor system has been challenged by several studies demonstrating that IOR exists as an output-, or output- plus input-based effect in simple keypress tasks not requiring oculomotor responses. Problematically, experiments in which keypress responses are required to visual events rarely use eye movement monitoring let alone the active discouragement of eye movement errors. Here, we return to an experimental method implemented by Ivanoff and Klein (2001) whose results demonstrated that IOR affected output-based processes when, ostensibly, only keypress responses occurred. Unlike Ivanoff and Klein, however, we assiduously monitor and discourage eye movements. We demonstrate that actively discouraging eye movements in keypress tasks changes the form of IOR from output- to input-based and, as such, we strongly encourage superior experimental control over or consideration of the contribution of eye movement activity in simple keypress tasks exploring IOR.

Averaging saccades are repelled by prior uninformative cues at both short and long intervals

When two spatially proximal stimuli are presented simultaneously, a first saccade is often directed to an intermediate location between the stimuli (averaging saccade). In an earlier study, Watanabe (2001) showed that, at a long cue–target onset asynchrony (CTOA; 600 ms), uninformative cues not only slowed saccadic response times (SRTs) to targets presented at the cued location in single target trials (inhibition of return, IOR), but also biased averaging saccades away from the cue in double target trials. The present study replicated Watanabes experimental task with a short CTOA (50 ms), as well as with mixed short (50 ms) and long (600 ms) CTOAs. In all conditions on double target trials, uninformative cues robustly biased averaging saccades away from cued locations. Although SRTs on single target trials were delayed at previously cued locations at both CTOAs when they were mixed, this delay was not observed in the blocked, short CTOA condition. We suggest that top-down factors, such as expectation and attentional control settings, may have asymmetric effects on the temporal and spatial dynamics of oculomotor processing.

Handles of manipulable objects attract covert visual attention: ERP evidence

Previous research has demonstrated that people are faster at making a manual response with the hand that is aligned with the handle of a manipulable object compared to its functional end. According to theories of embodied cognition (ETC), the presentation of a manipulable object automatically elicits sensorimotor simulations of the respective hand and these simulations facilitate the response. However, an alternative interpretation of these data is that handles preferentially attract visual attention, since attended stimuli and locations typically elicit faster responses. We investigated attentional biases elicited by manipulable and non-manipulable objects using event-related-potentials (ERPs). On each trial, a picture of a manipulable object was followed by a target dot that participants had to make a button-press to. The dot was located at either the handle or functional end of the object. Consistent with previous attentional cuing paradigms, we showed that the P1 ERP component was greater in response to targets cued by handles than by functional ends. These results suggest that object handles automatically bias covert attentional processes. These attentional biases may account for earlier behavioural findings, without any recourse to ETC.

Investigating a two causes theory of inhibition of return

It has recently been demonstrated that there are independent sensory and motor mechanisms underlying inhibition of return (IOR) when measured with oculomotor responses (Wang et al. in Exp Brain Res 218:441–453, 2012). However, these results are seemingly in conflict with previous empirical results which led to the proposal that there are two mutually exclusive flavors of IOR (Taylor and Klein in J Exp Psychol Hum Percept Perform 26:1639–1656, 2000). The observed differences in empirical results across these studies and the theoretical frameworks that were proposed based on the results are likely due to differences in the experimental designs. The current experiments establish that the existence of additive sensory and motor contributions to IOR do not depend on target type, repeated spatiotopic stimulation, attentional control settings, or a temporal gap between fixation offset and cue onset, when measured with saccadic responses. Furthermore, our experiments show that the motor mechanism proposed by Wang et al. in Exp Brain Res 218:441–453, (2012) is likely restricted to the oculomotor system, since the additivity effect does not carry over into the manual response modality.

In search of a reliable electrophysiological marker of oculomotor inhibition of return

Inhibition of return (IOR) operationalizes a behavioral phenomenon characterized by slower responding to cued, relative to uncued, targets. Two independent forms of IOR have been theorized: input-based IOR occurs when the oculomotor system is quiescent, while output-based IOR occurs when the oculomotor system is engaged. EEG studies forbidding eye movements have demonstrated that reductions of target-elicited P1 components are correlated with IOR magnitude, but when eye movements occur, P1 effects bear no relationship to behavior. We expand on this work by adapting the cueing paradigm and recording event-related potentials: IOR is caused by oculomotor responses to central arrows or peripheral onsets and measured by key presses to peripheral targets. Behavioral IOR is observed in both conditions, but P1 reductions are absent in the central arrow condition. By contrast, arrow and peripheral cues enhance Nd, especially over contralateral electrode sites.