Task difficulty is a major determinant of human performance. This statement might be considered as little more than a truism, but in fact it is a crucial issue when heterogeneous tasks and populations are considered. Unfortunately, it is all but a rare event to come across studies in which the mere difficulty imbalance between tasks or conditions can account for the observed differences in performance, without resorting to fine-grained explanations, which are instead often preferred. This problem is exacerbated in studies on brain-damaged patients, in which it might constitute a very serious flaw when groups are formed on the basis of performance in a given diagnostic task that is not matched for difficulty to the experimental task. This happens, for example, when two completely different tasks are used, one to select the patient group (or to rule out potential confounds) and one to collect the data of the study. The mere selection of patients for the presence of a given disorder often results in the selection of cases who are more cognitively impaired (the experimental group) versus cases who are less (the control group) (see Bonato et al. Reference Bonato, Sella, Berteletti and Umiltà2012b).
For these reasons, we are very sympathetic with the proposal of Kurzban et al., which aims at a better understanding of the relation between task difficulty and performance. We also strongly agree with their claim that the literatures on “self-control” (task difficulty, sustained attention, willingness to engage) and “executive functions” in cognitive neuropsychology/neuroscience are not sufficiently integrated. Several tasks classically described as loading executive functions can simply be construed as very difficult tasks; and it therefore seems reasonable to maintain that cognitive effort and executive functioning are largely overlapping concepts (Bonato et al. Reference Bonato, Priftis, Marenzi, Umiltà and Zorzi2012a), and do not depend on independent mechanisms, as strictly modularistic views would maintain. However, the opportunity cost model of Kurzban et al. appears too simple to account for the complex issue of the link between task difficulty and performance. Our skepticism revolves around two different lines.
First, the notion of a close relation between perceived effort and performance seems to take for granted that all cognitive processes are conscious. This is at odds with the widely accepted view that a large portion of our cognitive processes is not conscious (Bargh & Morsella Reference Bargh and Morsella2008; Sergent & Naccache Reference Sergent and Naccache2012), even in the case of very complex and apparently controlled tasks (e.g., for complex arithmetic, see Sklar et al. Reference Sklar, Levy, Goldstein, Mandel, Maril and Hassin2012). The mere notion that task performance can be driven by stimuli which are not consciously perceived suggests that performance and phenomenology might dissociate more often than in “rare pathological cases” (cf. target article, sect. 3.3, para. 7).
Second, several theoretical proposals show that task performance is closely dependent on the quantity and quality of the attentional load implied by the task. For instance, the load theory of attention (Lavie Reference Lavie1995; Reference Lavie2005) provides a comprehensive explanation of the influence of visual distractors, maintaining that their early or late filtering depends on the “load” of a concurrent attentional task, with reduced processing efficiency in the peripheral field (and hence less interference from distractors) when more attentional capacity is demanded by the central task. Higher perceptual load for central stimuli leads to exclusion of irrelevant peripheral inputs at an earlier stage, whereas under higher working memory load this exclusion occurs at a later stage. In other words, the processing efficiency seems to be related to the quantity and type of load, rather than to perceived subjective effort. Increased attentional load also reduces the will to spontaneously engage in internal, task-unrelated thoughts (Forster & Lavie Reference Forster and Lavie2009).
Stemming from this perspective, there is also robust evidence that, in right-hemisphere–damaged patients, the efficiency in contralesional hemispace processing is a function of the availability of attentional resources that can be engaged for monitoring visual space (Bonato Reference Bonato2012). In particular, patients show striking awareness deficits for the contralesional hemispace under multitasking in comparison to their baseline performance in a spatial monitoring task: The request to pay attention to an auditorily presented number or to a visual letter presented at fixation turns into inability to perceive targets appearing in the left hemispace (Bonato et al. Reference Bonato, Priftis, Marenzi, Umiltà and Zorzi2010). Thus, increasing task-difficulty caused by multi-tasking results in severe awareness deficits for the contralesional hemispace, regardless of the nature of the concurrent task (i.e., the “depleting task”; target article, sect. 3.1, para. 5). Indeed, patients showed the same severity of spatial awareness deficit regardless of whether they had to pay attention to visual or auditory channels while monitoring the visual space for target appearance (Bonato et al. Reference Bonato, Priftis, Umiltà and Zorzi2013). We maintain that the inability to perceive contralesional targets revealed with such an experimental procedure is the consequence of the impossibility to efficiently allocate attentional resources, which, in easier tasks, are allocated contralesionally and compensate for the patient's spatial deficits. Therefore, it seems that the classic view (Kahneman Reference Kahneman1973), according to which resources are not only dynamically allocated among tasks but also strictly depletable, constitutes the most economic explanation for these findings. This contrasts with the position of Navon (Reference Navon1984), who argued that the concept of resources is “unnecessary,” as well as with the view of Kurzban et al., who consider drops in performance as a consequence of individual trade-off between costs and opportunity.
More generally, the authors seem to maintain that the main reason for drops in performance can be traced back to a voluntary decision by the participant, which is hardly tenable in the case of our brain-damaged patients. Also, the idea that simultaneity and prioritization result in effective deployment of resources seems questionable in light of the patients' data. In summary, our results are in agreement with Kurzban et al.'s view that “task performance varies with the degree to which computational processes are allocated” and that resources can be “divided among multiple tasks” and “allocated in different portions to different tasks” (sect. 2.4.2, para. 1). Nevertheless, they also demonstrate that the “re-distribution” (or “reallocation”) of resources can be severely biased (spatially biased, in our case), and that it is not supported (at least in patients) by effective feedback mechanisms allowing the increase of attentional engagement when performance is unsatisfactory.
Task difficulty is a major determinant of human performance. This statement might be considered as little more than a truism, but in fact it is a crucial issue when heterogeneous tasks and populations are considered. Unfortunately, it is all but a rare event to come across studies in which the mere difficulty imbalance between tasks or conditions can account for the observed differences in performance, without resorting to fine-grained explanations, which are instead often preferred. This problem is exacerbated in studies on brain-damaged patients, in which it might constitute a very serious flaw when groups are formed on the basis of performance in a given diagnostic task that is not matched for difficulty to the experimental task. This happens, for example, when two completely different tasks are used, one to select the patient group (or to rule out potential confounds) and one to collect the data of the study. The mere selection of patients for the presence of a given disorder often results in the selection of cases who are more cognitively impaired (the experimental group) versus cases who are less (the control group) (see Bonato et al. Reference Bonato, Sella, Berteletti and Umiltà2012b).
For these reasons, we are very sympathetic with the proposal of Kurzban et al., which aims at a better understanding of the relation between task difficulty and performance. We also strongly agree with their claim that the literatures on “self-control” (task difficulty, sustained attention, willingness to engage) and “executive functions” in cognitive neuropsychology/neuroscience are not sufficiently integrated. Several tasks classically described as loading executive functions can simply be construed as very difficult tasks; and it therefore seems reasonable to maintain that cognitive effort and executive functioning are largely overlapping concepts (Bonato et al. Reference Bonato, Priftis, Marenzi, Umiltà and Zorzi2012a), and do not depend on independent mechanisms, as strictly modularistic views would maintain. However, the opportunity cost model of Kurzban et al. appears too simple to account for the complex issue of the link between task difficulty and performance. Our skepticism revolves around two different lines.
First, the notion of a close relation between perceived effort and performance seems to take for granted that all cognitive processes are conscious. This is at odds with the widely accepted view that a large portion of our cognitive processes is not conscious (Bargh & Morsella Reference Bargh and Morsella2008; Sergent & Naccache Reference Sergent and Naccache2012), even in the case of very complex and apparently controlled tasks (e.g., for complex arithmetic, see Sklar et al. Reference Sklar, Levy, Goldstein, Mandel, Maril and Hassin2012). The mere notion that task performance can be driven by stimuli which are not consciously perceived suggests that performance and phenomenology might dissociate more often than in “rare pathological cases” (cf. target article, sect. 3.3, para. 7).
Second, several theoretical proposals show that task performance is closely dependent on the quantity and quality of the attentional load implied by the task. For instance, the load theory of attention (Lavie Reference Lavie1995; Reference Lavie2005) provides a comprehensive explanation of the influence of visual distractors, maintaining that their early or late filtering depends on the “load” of a concurrent attentional task, with reduced processing efficiency in the peripheral field (and hence less interference from distractors) when more attentional capacity is demanded by the central task. Higher perceptual load for central stimuli leads to exclusion of irrelevant peripheral inputs at an earlier stage, whereas under higher working memory load this exclusion occurs at a later stage. In other words, the processing efficiency seems to be related to the quantity and type of load, rather than to perceived subjective effort. Increased attentional load also reduces the will to spontaneously engage in internal, task-unrelated thoughts (Forster & Lavie Reference Forster and Lavie2009).
Stemming from this perspective, there is also robust evidence that, in right-hemisphere–damaged patients, the efficiency in contralesional hemispace processing is a function of the availability of attentional resources that can be engaged for monitoring visual space (Bonato Reference Bonato2012). In particular, patients show striking awareness deficits for the contralesional hemispace under multitasking in comparison to their baseline performance in a spatial monitoring task: The request to pay attention to an auditorily presented number or to a visual letter presented at fixation turns into inability to perceive targets appearing in the left hemispace (Bonato et al. Reference Bonato, Priftis, Marenzi, Umiltà and Zorzi2010). Thus, increasing task-difficulty caused by multi-tasking results in severe awareness deficits for the contralesional hemispace, regardless of the nature of the concurrent task (i.e., the “depleting task”; target article, sect. 3.1, para. 5). Indeed, patients showed the same severity of spatial awareness deficit regardless of whether they had to pay attention to visual or auditory channels while monitoring the visual space for target appearance (Bonato et al. Reference Bonato, Priftis, Umiltà and Zorzi2013). We maintain that the inability to perceive contralesional targets revealed with such an experimental procedure is the consequence of the impossibility to efficiently allocate attentional resources, which, in easier tasks, are allocated contralesionally and compensate for the patient's spatial deficits. Therefore, it seems that the classic view (Kahneman Reference Kahneman1973), according to which resources are not only dynamically allocated among tasks but also strictly depletable, constitutes the most economic explanation for these findings. This contrasts with the position of Navon (Reference Navon1984), who argued that the concept of resources is “unnecessary,” as well as with the view of Kurzban et al., who consider drops in performance as a consequence of individual trade-off between costs and opportunity.
More generally, the authors seem to maintain that the main reason for drops in performance can be traced back to a voluntary decision by the participant, which is hardly tenable in the case of our brain-damaged patients. Also, the idea that simultaneity and prioritization result in effective deployment of resources seems questionable in light of the patients' data. In summary, our results are in agreement with Kurzban et al.'s view that “task performance varies with the degree to which computational processes are allocated” and that resources can be “divided among multiple tasks” and “allocated in different portions to different tasks” (sect. 2.4.2, para. 1). Nevertheless, they also demonstrate that the “re-distribution” (or “reallocation”) of resources can be severely biased (spatially biased, in our case), and that it is not supported (at least in patients) by effective feedback mechanisms allowing the increase of attentional engagement when performance is unsatisfactory.