2.1. Fundamentación Teórica
2.1.2. LA NORMATIVA PROCESAL PENAL
2.1.4.2. Muerte culposa
Theoretical accounts of the switch cost have fallen into two broad famOies. On the one hand. Allport and colleagues (e.g.. Allport et al., 1994; Allport & Wylie, 1999, 2000; Wylie & AUport, 2000) have suggested that switch costs reflect an interference effect caused by the carryover of a previous task set into switch trials. Such hypotheses will be referred to as 'task carryover' accounts. For example, the version of the task carryover account put forward by Allport et al. (1994) described this carryover as 'task set inertia' (TSI). According to this hypothesis, a task set active on one trial will persist, involuntarily, on the next. As well as a carryover of the previously active task set on switch trials. Allport et al. (1994) hypothesise that inhibition of competing task sets can also persist on switch trials, when the previously competing task set is now required. Thus, various task sets may be primed, positively or negatively, depending on the task performed on the previous trial. On a switch trial, the required task set may be negatively primed, and an inappropriate task set may be positively primed. As a result, responses may be slower and more error-prone, hence the switch cost. According to this account, in common with otirer task carryover accounts, there is no need to posit differences in the higher-level cognitive processes that take place on switch and nonswitch trials. Rather, one or more of the processes that take place on nonswitch trials may be prolonged by competition on switch trials. Typically, it is assumed that the process most likely to be prolonged in this way is response selection.
Task carryover accounts, such as the TSI hypothesis, may be contrasted with 'extra process' accounts (e.g. Monsell, Yeung, & Azuma, 2000; Rogers & Monsell, 1995; see also De Jong, 2000; Hiibner, Futterer, & Steinhauser, 2001; Rubinstein et al., 2001). According to these accounts, switch costs reflect the duration of one or
more stage-like control processes that reconfigure the cognitive system for the upcoming task. For example, in the version of the extra process account put forward by Rogers and Monsell (1995), there are two extra control processes that must take place on switch trials, but do not occur on nonswitch trials. One process cannot be completed until the arrival of the first stimulus of the new task (i.e. it is a stimulus-driven ^exogenous control process'). An additional component of the switch cost is hypothesised by Rogers and Monsell (1995) to reflect the operation of an 'endogenous control process' that can be executed before the arrival of the first stimulus in the new task.
It is important to have a clear definition of exactly what separates task carryover and extra process accounts. It is clearly not an adequate characterisation of task carryover accounts to say merely that the switch cost arises from the effects of the task performed on the previous trial. Since the only difference between switch and nonswitch trials is the task performed on the previous trial, the distinction between the two types of account would collapse under this characterisation. The definition of extra process accounts is fairly straightforward: they claim that the switch cost measures the duration of one or more cognitive processes that take place on switch trials but do not take place on nonswitch trials. Task carryover accounts will be defined here as the negation of this. They claim that there is no need to posit additional cognitive processes on switch trials in order to explain the switch cost, either because no such additional cognitive processes exist, or because the duration of such processes is not typically what is being measured by the switch cost. According to such accounts, the switch cost reflects an interference effect (i.e. a form of priming), rather than measuring the duration of any control process.
Although task carryover accounts have been characterised as the negation of extra process accounts, the two types of account need not be mutually
incompatible. It is of course possible that switch trials are slowed both by one or more additional processes (in line with extra process accounts) and a priming effect, leading to slowed execution of the processes that normally take place on nonswitch trials (in line with task carryover accounts). One such hybrid account has been presented by Meiran (2000a, 2000b). Evidence for extra process or task carryover accounts does not, therefore, logically exclude the other type of account. However, if it could be shown that task carryover or extra process accounts are able to provide an adequate explanation of task switching data on their own, this would make the other sort of account unnecessary. The assumption of extra processes on switch trials, for example, would not serve any explanatory role if it could be shown that a task carryover effect was able to explain all of the data relevant to the debate between the two sorts of account^. In this limited sense, then, the two types of account can be seen as competing. The various types of account are illustrated in Figure 2.1, adapted from Monsell et al. (2000, Figure 2).
These accounts differ in their imptications for the relationship between task switching performance and executive function. According to extra process accounts there is a relatively straightforward relationship between control processes and the switch cost. Since switch trials are extended by the duration of one or more stage-like control processes, the RT increase on switch trials can be assumed to reflect the duration of these processes, independent of the lower- level processes involved in task performance. Some authors have claimed that if the switch cost simply reflects a priming effect, as proposed by task carryover accounts, it would therefore be irrelevant to the study of control processes. For example, Rubinstein et al. (2001) describe Allport et al.'s (1994) TSI hypothesis as
^ In order to prefer a task carryover over an extra process account in this circum stance, it w ould also be necessary to show that there is som e data that can be m ore adequately explained by the task carryover account. O therw ise, the tw o types o f account w ould be equally adequate.
Nonswitch trial
Stimulus
onset Response
Basic S-R
Switch trial
o) Extra process account
Stimulus onset
Switch
cost
Response Extra process Basic S-R processesbj Tasir carryover account
Stimulus onset Response Basic S-R processes (e x te n d e d by com petition)
cj Hybrid account
Stimulus onset Response Basic S-R processes (e x te n d e d by com petition)Figure 2.1 Theoretical accounts of the RT switch cost. 'Basic S-R processes' refer to the processes that take place on nonswitch trials. According to extra process accounts, one or more extra processes slow reaction time on switch trials, while the duration o f the basic S-R processes is unchanged. Task carryover accounts attribute the switch cost to extended basic S-R processes on switch trials.
attributing switch costs to "conflicts at a basic level of task processing without any assumption of control being exerted from a higher supervisory level" (p. 790). However, task carryover accounts need not rule out the involvement of control processes in task switching ("Undoubtedly, additional control processes are brought to bear on the pre-existing task-set, on a ""switch"" trial. If not, the previous task would continue to be executed"; Allport & WyUe, 1999, p. 280). Such accounts merely deny that these control processes are measured in any direct way by the switch cost. Information about control processes might nonetheless be inferred from task switching data. Thus, even according to task carryover accounts, the task switching paradigm may be of relevance to the study of executive function. However, according to such accounts the relationship between control processes and switch costs is indirect.