Perceptual competition between targets and distractors determines working memory access and produces intrusion errors in rapid serial visual presentation (RSVP) tasks.

Abstract

When a target and a distractor that share the same response dimension appear in rapid succession, participants often erroneously report the distractor instead of the target. Using behavioral and electrophysiological measures, we examined whether these intrusion errors occur because the target is often not encoded in working memory (WM) or are generated at later postencoding stages. In 4 experiments, participants either provided two guesses about the target’s identity, or had to select the target among items that did not include the potential intruder. Results showed that the target did not gain access to WM on a substantial number of trials where the distractor was encoded. This was also confirmed with an electrophysiological marker of WM storage (CDA component). These findings are inconsistent with postencoding accounts of distractor intrusions, which postulate that competitive interactions within WM impair awareness of the target, the precision of target representations, or result in the target being dropped from WM. They show instead that target-distractor competition already operates at earlier perceptual stages, and reduces the likelihood that the target gains access to WM. We provide a theoretical framework to explain these findings and how they challenge contemporary models of temporal attention.

Grand Challenges in Consciousness Research Across Perception, Cognition, Self, and Emotion

“…However, the distinction between phenomenal and access consciousness and the overflow argument appear supported by transient vs. sustained recurrent mechanisms in the cerebral cortex (Lamme, 2003;Raffone and Pantani, 2010;Raffone et al, 2014), which are consistent with a global workspace theoretical framework, and can account for large sets of findings with attentional blink and visuo-spatial working memory paradigms (Simione et al, 2012). Moreover, attentional processes can operate at different stages, with differential influences on perceptual (phenomenal) consciousness and conscious (cognitive) access (Lamme, 2003;Raffone et al, 2014;Simione et al, 2020;Zivony and Eimer, 2020). Finally, the global workspace theoretical framework needs to take into account developments on brain networks (e.g., Bressler and Menon, 2010), that appear consistent with large-scale broadcasting in the brain before the stage of conscious access, which may support phenomenal consciousness.…”

The diachronic account of attentional selectivity

“…Overall, these behavioural findings suggest that competitive interactions between items within the same attentional episode can affect the ability of any of these items to be encoded within WM and become accessible to perceptual reports. In a recent study (Zivony and Eimer, 2020), we provided more direct behavioural and electrophysiological evidence for this claim. In a task with two lateral RSVP streams, participants were allowed two guesses about the target's identity.…”

“…In two-target RSVP attentional blink tasks, the lag-1 sparing effect is assumed to occur because T1 and T2 are both encoded within the same attentional episode. Electrophysiological evidence for this comes from observations that often only a single N2pc component (a marker of attentional engagement, see above) is elicited when T1 and T2 appear in immediate succession (Tan and Wyble, 2015: see also Akyürek & Meijerink, 2012;Callahan-Flintoft, Chen, & Wyble, 2018;Zivony & Eimer, 2020). This suggests that the attentional engagement triggered by T1 is sufficient for both targets to gain access to WM.…”

“…In visual search studies, target N2pcs not only emerge earlier for participants who show fast responses to targets than for those with slower responses, but also within participants for trials with fast versus slow response times (Drisdelle, West, & Jolicoeur, 2016; see also Eimer & Mazza, 2005, for similar N2pc onset differences in a change detection task). In RSVP tasks, trials with post-target distractor intrusion errors are associated with delayed N2pc onsets relative to trials with correct responses (Zivony & Eimer, 2020;. This N2pc onset difference remains even when attention is focused in advance on the target RSVP stream (Zivony & Eimer, 2021, Experiment 3), demonstrating that it does not reflect variability in the speed of attentional shifts.…”

“…Variations of this paradigm produce two highly reliable patterns of results. First, when the distractors share the target's response dimension (e.g., when the target is a digit in a stream of other digits, Figure 2B), distractor intrusions are frequent: Participants will often erroneously report not the target, but the identity of a distractor that appears in close temporal proximity to the target (Botella, Suero, & Barriopedro, 2001;Chun, 1997;Goodbourn & Holcombe, 2015;Vul et al, 2008;Zivony & Eimer, 2020;. Second, when participants are asked to identify two successive targets (T1 and T2) in the RSVP stream (Figure 2C), the accuracy of reporting both targets is high when they appear more than half a second apart, but report accuracy for T2 is strongly impaired when it is presented within 200ms-500ms after T1 (attentional blink, Raymond, Shapiro, & Arnell, 1992; see Zivony & Lamy, 2020, for a discussion of different accounts of this phenomenon).…”