Dr. Rosenholtz suggests that many attention-attributed phenomena in the visuospatial domain can be accounted for by visual crowding confounds, concluding that perception is the factor limiting performance rather than attention. We support this attempt to banish attention as an explanatory mechanism in the “spatial attention” and “object attention” literatures and think this will lead to a more accurate understanding of these phenomena (see also Di Lollo, Reference Di Lollo2018; Hommel et al., Reference Hommel, Chapman, Cisek, Neyedli, Song and Welsh2019). We further suggest that perceptual dynamics may account for findings in the “temporal attention” literature. At its core, the present “crisis” in visual attention research is an issue of circular causal inference – attention is assumed to be the cause of improved performance because performance improves. Examples of this can be found in the study of where attention operates (Awh, Belopolsky & Theeuwes, Reference Awh, Belopolsky and Theeuwes2012; Posner, Reference Posner1980), and when attention operates (Cherry, Reference Cherry1953; Treisman, Reference Treisman1964). Thus, if there is merit to banishing attention from the study of spatial attention, it should also be banished from the study of temporal attention.

Temporal attention is the selection of target stimuli for report (e.g., report “target” letters while ignoring “distractor” digits) when stimuli are presented over a brief period of time. It is often studied with the rapid serial visual presentation (RSVP) procedure in which many stimuli are presented in the same spatial location at rates faster than eight per second. Although people can easily report one target in an RSVP stream, they often fail to report a subsequent target presented shortly after the first (Broadbent & Broadbent, Reference Broadbent and Broadbent1987). Because the report of the later-occurring target is preserved when participants are instructed to ignore the first target, the phenomenon was deemed an attentional deficit and so named the “attentional blink” (AB: Raymond, Shapiro & Arnell, Reference Raymond, Shapiro and Arnell1992). This terminology reflects the common assumption that limited attentional resources are busy encoding the first target and therefore, those attentional resources cannot be deployed to a second target. In the ensuing decades, the AB literature has blossomed, revealing a highly complex set of auxiliary data patterns such as “lag 1 sparing,” “the spread of sparing,” and a temporal shift of intrusions with lag, to name just a few highly reliable effects. To account for this ever-growing list of effects, sophisticated mathematical models of temporal attention have been developed and refined, with many new theoretical variants including additional mechanisms to accommodate additional results. This is the way that science should progress, but we suggest that a simpler account of the AB literature might emerge from the hypothesis that high-level perceptual dynamics explain the failure to report a second target.

In keeping with Rosenholtz’s suggestion to remove attention as an explanation, we refer to the basic “Attentional Blink effect” as a “second target deficit” to avoid implying a particular explanation. Rosenholtz proposes that visual crowding is the principal factor limiting performance in visual-spatial tasks, and the analog in the RSVP task is “temporal crowding” (i.e., competing items presented immediately before or after a target stimulus). Temporal crowding is measured by the inter-stimulus-interval (ISI) between the offset of one stimulus and the onset of the next. For instance, the magnitude of the second target deficit varies both as a function of the interval between the first and second target as well as the ISI between intervening distractors and the targets (e.g., Chen & Zhou, Reference Chen and Zhou2015). The role of temporal crowding (e.g., “visual masking”) in second target deficits has been the topic of considerable debate; however, it is often discussed as interacting with temporal attention (Brehaut, Enns, & Di Lollo, Reference Brehaut, Enns and Di Lollo1999; Grandison, Ghirardelli, & Egeth, Reference Grandison, Ghirardelli and Egeth1997), rather than the cause of the second target deficit.

Another form of temporal crowding is termed “repetition blindness” (Kanwisher, Reference Kanwisher1987), which is a transient inability to appreciate that a second occurrence of an item is a unique perceptual event. We refer to this as a transient repetition deficit rather than repetition blindness, considering that we cannot be certain whether this is a perceptual effect. Of note, the repetition deficit with letter stimuli exhibits a time course that is nearly identical to the second target deficit in the AB literature (Rusconi & Huber, Reference Rusconi and Huber2018). Transient repetition deficits are observed for a wide variety of stimuli and tasks, including high-level semantic categories (Tian & Huber, Reference Tian and Huber2013), as expected if they reflect neural dynamics at all levels of perceptual/conceptual processing (Huber, Reference Huber and Ross2014). Thus, the second target deficit in the AB literature might reflect a failure to perceive that the second target belongs to the category of target items. For instance, in an RSVP task with two different target letters that occur in a stream of distracting digits, the observer might perceive the line segments and visual primitives of the second target but fail to perceive that the second target is a letter. This is the core assumption of the “perceptual wink” model of Rusconi and Huber (Reference Rusconi and Huber2018), which explains both transient repetition deficits and second target deficits. In short, it is not clear whether one needs to hypothesize a complex temporal attention system when considering the role of perceptual dynamics.

The cause of the second target deficit in the AB literature is the subject of ongoing debate, and there might not be a definitive answer to this question until such time as we have neuroimaging techniques with millisecond-level measurement of all perceptual responses. In the meantime, we should follow Dr. Rosenholtz’s lead in attempting to explain as much as we can by applying well-understood perceptual mechanisms to the study of attention. Perceptual mechanisms such as visual crowding or temporal crowding cannot be “turned off” just because a study is designed to study attention rather than perception. If these perceptual mechanisms occur in studies of perception, they will also occur in studies of attention. At a minimum, we should be careful in assigning labels that entail a theoretical connotation, such as “attentional blink” or “repetition blindness.” Instead, labels should simply describe the effect of interest – E.g., a second target deficit or a repetition deficit.