Understanding the capacity limit of how much can be processed at a given time is at the core of understanding visual attention. Rosenholtz argues that existing methodologies cannot clearly identify a unifying capacity limit and instead proposes that a unifying limit may exist in task complexity. Although Rosenholtz’s proposal to consider task complexity could be promising, (1) it lacks a precise definition of task complexity and (2) it ignores the internal state of the observer and incorrectly assumes that the effort an observer is able or willing to invest is constant. We argue that pursuing task complexity as a solution to the proposed problem necessitates the consideration of the amount of effort an observer exerts.

Rosenholtz uses the example of a categorical classification task to describe task complexity (e.g. “Is this a coffee table or dining room table?”). Rosenholtz proposes that the complexity of the decision boundary may reflect a given task’s complexity. This definition is problematic for two reasons. First, due to the strict focus on categorical classification tasks, it is unclear how the definition precisely extends to other visual tasks. Second, this definition does not allow one to clearly measure what a decision boundary actually looks like in a given task, making it difficult to test the proposal empirically.

How to quantify task complexity? Behavioral outcomes such as response accuracy and speed may be informative of task complexity. However, as Rosenholtz points out, task complexity cannot be equated to task difficulty when based purely on behavioral outcomes. One key factor to consider is the amount of effort observers invest in a task. Task difficulty and cognitive effort are related concepts but are certainly separable (Shenhav et al., Reference Shenhav, Musslick, Lieder, Kool, Griffiths, Cohen and Botvinick2017; Westbrook & Braver, Reference Westbrook and Braver2015). For example, performance on a task with constant difficulty is modulated strongly by the amount of effort the observer is willing to exert. Put differently, if an observer is ‘trying harder’ on a given task, capacity may appear high and vice versa. Consequently, the complexity of any task cannot solely be quantified by task performance in isolation, and the effort observers invest in the task must be incorporated.

We highlight two ways that effort may modulate capacity estimates. First, the amount of effort an observer is willing/able to exert fluctuates across trials and conditions and also differs across individuals. Fortunately, the amount of effort an observer exerts can be reliably tracked using pupil size (Beatty, Reference Beatty1982; Bumke, Reference Bumke1911; Kahneman, Reference Kahneman1973; Koevoet, Strauch, et al., Reference Koevoet, Strauch, Van der Stigchel, Mathôt and Naber2024; Strauch et al., Reference Strauch, Wang, Einhäuser, Van der Stigchel and Naber2022; van der Wel & van Steenbergen, Reference van der Wel and van Steenbergen2018). For example, pupil size sensitively tracks how much or how precisely material is held in working memory (Master, Li, & Curtis, Reference Master, Li and Curtis2024; Robison & Unsworth, Reference Robison and Unsworth2019; Unsworth & Miller, Reference Unsworth and Miller2021; Unsworth & Robison, Reference Unsworth and Robison2017). Moreover, fluctuations in pupillary responses across trials reliably predict performance (Koevoet, Naber, et al., Reference Koevoet, Naber, Strauch, Somai and Van der Stigchel2023, Reference Koevoet, Naber, Strauch and Van der Stigchel2024; Koevoet, Strauch, et al., Reference Koevoet, Strauch, Van der Stigchel, Mathôt and Naber2024; Robison & Unsworth, Reference Robison and Unsworth2019; Unsworth & Miller, Reference Unsworth and Miller2021). Pupil size even captures individual differences in the effort invested or necessary to solve a task (Ahern & Beatty, Reference Ahern and Beatty1979; Robison et al., Reference Robison, Ralph, Gondoli, Torres, Campbell, Brewer and Gibson2023; Robison & Brewer, Reference Robison and Brewer2020; Unsworth & Robison, Reference Unsworth and Robison2017). These findings indicate that pupil size is a sensitive measure of exerted effort in a given task across trials, conditions, and observers and show that effort is linked but cannot be equated to task difficulty. Potentially, a combination of the exerted effort with behavioral outcomes may inform task complexity.

Second, observers tend to prefer less effortful ways to solve tasks. When aiming to identify a putative unifying capacity limit, experimenters would need observers to “try as hard as possible” to circumvent underestimations of capacity. A recent accumulating body of work demonstrates that when given a free choice, observers generally prefer to solve (visual) tasks while exerting the least amount of effort as possible (Ballard, Hayhoe, & Pelz, Reference Ballard, Hayhoe and Pelz1995; Böing et al., Reference Böing, Ten Brink, Hoogerbrugge, Oudman, Postma, Nijboer and Van der Stigchel2023; Draschkow, Kallmayer, & Nobre, Reference Draschkow, Kallmayer and Nobre2021; Hoogerbrugge et al., Reference Hoogerbrugge, Strauch, Nijboer and Van der Stigchel2023; Sahakian et al., Reference Sahakian, Gayet, Paffen and Van der Stigchel2023; Somai, Schut, & Van der Stigchel, Reference Somai, Schut and Van der Stigchel2020). One example of such behavior comes from the copying task wherein observers copy a pattern of multiple objects. Crucially, observers are free to reinspect the example pattern as often of they would like. Despite the fact that observers could likely encode ∼3 items every time they inspect the example grid (Luck & Vogel, Reference Luck and Vogel1997), they generally choose to encode only a single item at a time and look back each time after placing an item. Observers likely weigh the required effort necessary to load multiple items against the relatively subtle cost of reinspecting the example grid by making an eye movement (Ballard et al., Reference Ballard, Hayhoe and Pelz1995; Van der Stigchel, Reference Van der Stigchel2020). Indeed, observers prefer to choose less effortful ways to solve tasks whenever possible.

This principle even extends to the level of eye movements. We recently demonstrated that pupil size can even measure the effort of planning and executing saccades across directions (Koevoet, Strauch, et al., Reference Koevoet, Strauch, Naber and Van der Stigchel2023). We then let observers freely choose between two saccade targets in different directions. Results showed that participants preferred saccading in directions associated with less effort (Koevoet, Van Zantwijk, et al., Reference Koevoet, Van Zantwijk, Naber, Mathôt, Van der Stigchel and Strauch2024). Put differently, observers prefer the “lazy” option even at the level of eye movements (see Hoppe & Rothkopf, Reference Hoppe and Rothkopf2016; Kadner et al., Reference Kadner, Thomas, Hoppe and Rothkopf2022; Thomas, Hoppe, & Rothkopf, Reference Thomas, Hoppe and Rothkopf2022). Thus, a crucial consideration when incorporating effort into the quantification of a putative unifying limit on visual processing must also consider that observers tend to avoid effort exertion whenever possible.

Rosenholtz proposes that identifying a putative unifying capacity limit may be achieved by considering task complexity. The quantification of task complexity may be possible by combining behavioral outcomes and the amount of exerted effort. Pupil size could serve as a potent tool to track differences in effort exertion across trials, conditions, and observers. Moreover, tasks should be designed that make observers “try as hard as possible” to prevent underestimations of capacity limits. Together, if task complexity may be used to identify a unifying capacity limit, effort must be considered to prevent an impending crisis for the field of visual attention.