LA ESTRUCTURA ORGANIZACIONAL DEL ÁREA COMERCIAL

Although both children and adults may engage in common information processes in switching between tasks, it is not clear what measures and what measurement methods reflect developmental changes best. The common measures used in task-switching studies are between-condition mixing cost and switch costs in RT and accuracy, as well as overall RT and accuracy on different trial types (pure trials in single-task block, repetition and switch trials in mixed-task block). In what follows, I will discuss these measures in the few developmental studies that have employed procedures similar to those in the adult task-switching studies.

1.4.2.1 Mixing cost

Several developmental studies that have employed a block design with pure- and mixed-task blocks have found developmental effects on mixing cost.

For example, Davidson et al. (2006) carried out a series of experiments, involving task-switching elements, with children aged between 4 and 13 years of age, as well as on adults. The tasks involved response conflicts with spatial incompatibility. These required the participants to switch between responding on either the same or the opposite side of the screen to where the target stimulus appeared, based on the form of the stimulus. The authors reported greater global mixing costs in accuracy for children aged 10 years or younger, than for older participants. However, the accuracy mixing costs appeared as the result of a trade-off with RTs, with older participants showing greater RT mixing costs than younger children. Although accuracy mixing costs were greater with younger participants, this trade-off effect renders the overall effect of age on mixing costs difficult to interpret.

In another study with young children, Dibbets and Jolles (2006) carefully designed tasks suitable for preschool children (i.e. tasks on which young children had high accuracy). They also reported greater accuracy mixing costs among the youngest participants (aged from 4.8 years to 13 years of age), but no greater RT mixing costs. Thus, there was no change in the cost criteria between age groups (i.e., no speed-accuracy trade-off). However, some other studies have found an age effect on RT and/or accuracy mixing costs (Cepeda et al., 2001, with participants aged 7 to 82 years of age; Kray, Eber, & Karbach, 2008, with participants aged 7 to 77 years of age; Reimers & Maylor, 2005, with participants aged 10 to 66 years of age). Of course, these latter studies involved children older than the preschool years. Overall, past research generally

supports the presence of an age effect on mixing costs, although the types of costs (i.e. RT or accuracy) are less consistent across studies. Finally, at least one experiment has failed to find any influence of age on mixing costs using a colour/shape choice task, despite a large age gap in participants (7-year-olds vs. University students [Exp. 1], Ellefson, Shapiro, & Chater, 2006). Thus, this raises the question of whether the effect age on mixing costs is robust across different types of tasks.

1.4.2.2 Switch cost

Switch costs were found to interact with age in some studies (Cepeda et al., 2001, [7 to 82 years old]; Chevalier, Martis, Curran, & Munakata, 2015, [5 and 10 years old]; Crone, Bunge, Van Der Molen, & Ridderinkhof, 2006, [7 to 25 years old]; Davidson et al., 2006, [4 to 13 years old and adults]). However, numerous other studies have found no age effect on either RT or accuracy switch costs (e.g. Dibbets & Jolles, 2006, [4.8 to 13 years old]; Ellefson et al., 2006, [7 years old and adults]; Reimers & Maylor, 2005 [10 to 66 years old]). If switch cost is an effective measure of cognitive control in task set reconfiguration, and/or goal maintenance, developmental differences in switch costs should be observed. However, as discussed previously, switch costs can also be the result of task set inertia, which may or may not be resolved with control mechanism such as inhibitory control. If inhibitory control is needed to resolve effects of task set inertia, then younger children may exhibit greater switch costs than older participants. However, if task set inertia is just the result of the information processes, irrespective of age, then switch cost may not exhibit reliable interaction with age. Thus, at the moment it remains unclear whether switch costs are a meaningful correlate of cognitive development.

1.4.2.3 Processing speed

Many developmental studies suggest that processing speed changes as a function of global development, and is often seen as a mental resource that mediates other general ability such as IQ and fluid intelligence (Robert Kail, 2000; Park, Mainela-arnold, & Miller, 2015). This global account of processing speed comes from the observation that reaction time variability across different tasks can be mostly captured by simple linear regressions (Hale, 1990; Robert Kail, 2000; Kiselev, Espy, & Sheffield, 2009; Miller & Vernon, 1997). However, while much of the developmental difference in speeded tasks may be accounted for by processing speed alone, merely looking at processing speed in isolation may run the risk of overlooking other potentially age-dependent, function-specific and task-specific processes in complex speeded tasks. The issue of how changes in function-specific and task-specific processes may interact with the global development indexed by processing speed is particularly pertinent when it comes to young children between 3 to 6 years of age, since a slight change in task parameter can often have a dramatic impact on task performance (Huizinga, Dolan, & van der Molen, 2006). It is not clear, for example, if components in the task set inertia (e.g. priming cost) are more or less easily overcome because of differences in global processing speed, or because of some other specific functional processes such as working memory and inhibitory control.

1.4.2.4 Errors

Developmental studies of cognitive flexibility have shown that different types of errors decrease with age—perseverative errors or errors caused by difficulty in shifting goals (Crone, Ridderinkhof, Worm, Somsen, & van der Molen, 2004; Zelazo, 2006), distraction errors caused by weak goal

maintenance (Carroll, Blakey, & FitzGibbon, 2016; Chevalier & Blaye, 2008;

Crone et al., 2004), and errors caused by prepotent motoric responses (Ling, Wong, & Diamond, 2016; Wright & Diamond, 2014). Although all error types decrease as a function of age, different error types may have different developmental trajectories. Crone and colleagues (2004) looked into this issue with participants from 8 to 25 years of age, on a task structurally similar to Wisconsin Card Sorting Task (WCST), and using a task-switching paradigm where the participants had to change rule through induction with error feedback or with explicit cues. They measured perseverative error on the trials immediately after the error feedback, and the distraction error on trials where the sorting behaviour changed from the chosen rule on the previous trial (i.e.

without error feedback or being instructed so). The found that adult-levels of perseverative error were reached earlier in development than adult-levels of distraction error (Crone et al., 2004). This result was interpreted as evidence of an earlier maturation of the ability to shift goals than the ability to actively maintain task set.

Since the age group of interest in the current thesis is much younger than in Crone et al.’s study, it is likely that the errors committed by young children would encompass all error types. Thus, in the current thesis, instead of differentiating error types, accuracy serves only as an indicator of the overall level of development than to the development of a specific cognitive function.

To achieve an overview of development by accuracy measure, it is important to have an experimental design that mitigates biases for a specific error type.

Perseverative error happens only when the pre-switch rule can also be applied to the post-switch stimulus, such as when the stimulus is bivalent. In comparison, distractor errors may happen more sporadically, although they may

still be tied to the goal representations (Blakey & Carroll, 2018). It is possible that by intermixing bivalent, univalent and neutral stimuli in a task set, biases towards a specific error type can be minimised.

1.5 Flexible goal-directed behaviours in a multisensory