Ruth Rosenholtz trenchantly argues that there is a Kuhnian crisis regarding visual attention. Yet is there a paradigm, an overarching theory of attention, such that it faces anomalies? I’m unsure. For example, the Treismanian framework (target article, Figure 1) does not figure centrally in the neuroscience of visual search with nonhuman primates. Arguably, there is no paradigm that everyone agrees with, and absent a unifying account, we find fragmentation as exemplified in the proliferation of postulated mechanisms said to be attention (Rosenholtz, target, p. 3). Disunity is the crisis.

Disunity leads to skepticism: there is no such thing as attention and no one knows what attention is (B. Anderson, Reference Anderson2011; Hommel et al., Reference Hommel, Chapman, Cisek, Neyedli, Song and Welsh2019). Yet this crisis is self-imposed due to our failure to recognize a shared methodological commitment to a functional conception of attention (Wu, Reference Wu2024). This commitment, which provides the functional structure of attentional tasks, reflects attention as the folk know it. In everyday actions, when a person mentally selects something to respond to, they attend to it (animals too, Hommel et al. op. cit.). So, we visually attend to a target when we visually select it to inform behavior. In visually selecting something to remember, to grasp, to avoid, to report, or to ponder, we visually attend to it to respond appropriately.

This common conception of ordinary attention engenders the shared methodological commitment. To study visual attention, we design appropriate tasks that require specific visual selection for correct performance. Selection is set by task instructions that manipulate how subjects attend. Subjects perform experimental tasks akin to mundane actions, so science studies attention as we all know it. Rosenholtz is right that “selection and performing a task might be inseparable” (24). They are inseparable. Accordingly, visual attention as what guides a person’s doing something is what we’re trying to explain.

The science of attention is anchored on experimental tasks built from the common conception. Consequently, the fragmentation of attention into distinct, often underspecified, mechanisms that rightly bothers Rosenholtz amounts to science changing the topic from the shared conception found in the folk and (methodologically) among scientists. Neural mechanisms are a different thing. They are not attention but explain it. Worse, this fragmentation of attention into distinct internal mechanisms, when not due to careless labeling, results from bad logic: reasoning from a process’s being selective to concluding that it is attention. Because many selective processes are not attention, the inference is a non sequitur, the conclusion a category mistake.

This allows us to recast Rosenholtz’s useful exercise of banning attention since her emphasis on eliminating the word ‘attention’ obscures its force. The issue is not terminology but explanation and ontology. After all, if “attention” successfully refers to a mechanism, nothing is gained by treating it as “that which should not be named”. Rather, her exercise aims at explanatory parsimony. We do without attention by showing that the referenced mechanism is not needed to explain performance. For example, the spotlight of attention as (a stand in for) a mechanism to explain behavioral effects in spatial cueing can be eliminated in favor of appealing to expectations (statistical beliefs) and task representations (intentions) (Wu, Reference Wu2025, Chapter 4).

These points bear on Rosenholtz’s discussion of inattentional blindness as a sign of crisis. Inattentional blindness informs debates about phenomenal consciousness (a mistake, see (Wu, Reference Wu2025, Chapter 10)), but the basic point consistent with the data is this: when attention is not directed at a target, subjects will fail to access it for memory or report. Attention, as what the subject does (“focus”), is necessary for access. It gates. Inattentional blindness tests this by experimentally eliminating attention to salient stimuli, say a gorilla, and recording failure to report it. This deploys the shared methodological commitment: attention is directed away from the gorilla by requiring subjects to visually select a distinct target, a ball, for a counting task.

Rosenholtz suggests that to “rethink attention, we should try to describe phenomena in a more agnostic way that does not rely on earlier concepts of attention or selection” (target article, 12). I agree if attention is misidentified with internal mechanisms (cf. her earlier mechanistic suggestion that “attention could narrow the pooling mechanisms that lead to crowding”, target article, 23). Yet inattentional blindness fundamentally concerns selective attention in the ordinary sense. Indeed, we learn the basic point at a young age when teachers admonish us that in focusing on one thing (Tik-Tok), we miss other things (the lesson). One hardly needs experiments to establish this.

What needs rethinking is not whether inattentional blindness concerns attention (it does), but its significance. Such paradigms probe distractability during performance, and distraction, as William James noted, is essentially contrasted with (task-directed) attention. Further, we can explain distractability without appeal to spotlights (attention as mechanism). For example, we can deploy priority maps that integrate various biases including task representations (goals), expectation, value, saliency maps and other factors. Attention as pulled from task is explained by the influence of various biases such as value (B. A. Anderson et al., Reference Anderson, Kim, Kim, Liao, Mrkonja, Clement and Grégoire2021). No attentional mechanism (a “neural spotlight”) is needed as priority maps and historical factors might suffice, and a priority map is not attention because it explains the subject’s (shifts of) attention (Bisley & Mirpour, Reference Bisley and Mirpour2019).

Despite agreement on “banning”, I disagree with Rosenholtz concerning the main crisis. A healthy science clashes about mechanistic explanations of core phenomena such as search. Her commentators will likely debate her crossing out moves depicted in (target article, Figure 5). This is “normal” science anchored on a paradigmatic experimental task. The real crisis is the continued proliferation of attention into underspecified mechanisms and processes, leading to fallacies, confusion, skepticism and pessimism. This fragmentation blocks us from seeing that a common conception of attention unifies decades of empirical work (Wu, Reference Wu2025, Chapters 4–5). We should eliminate attention as an internal mechanism. It is not (B. Anderson, Reference Anderson2011; Desimone & Duncan, Reference Desimone and Duncan1995; Hommel et al., Reference Hommel, Chapman, Cisek, Neyedli, Song and Welsh2019; Krauzlis et al., Reference Krauzlis, Bollimunta, Arcizet and Wang2014; Wu, Reference Wu2023). Attention is an activity we undertake as part of doing things in the world.