Derechos de las víctimas

Evidence for task carryover accounts has come from findings indicating an effect of performance of an earlier task on subsequent task switching performance (e.g. Allport et al., 1994; Allport & Wylie, 1999, 2000; Mayr & Keele, 2000; Waszak, Hommel, & Allport, in press; Wylie & Allport, 2000).

2.6.1 Asymmetric switch costs

Allport et al. (1994, Experiment 5) carried out an experiment where subjects switched between word reading and colour naming in response to incongruent Stroop stimuli. As expected, word reading yielded faster reaction times than colour naming (see MacLeod, 1991). However, this experiment also produced a most unexpected result. Reaction times for word-reading trials were slower when colour naming had been performed on the previous trial (i.e., there was a switch cost), but AUport et al. failed to detect a switch cost for colour-naming

trials that followed performance of the word-reading task. In other words, the switch cost appeared to be confined to the switch from the nondominant into the dominant (i.e., better learned, easier) task.

Subsequent experiments replicated this asymmetry in switch costs (Allport & Wylie, 2000; Wylie & Allport, 2000). AUport and Wylie found that there is a cost for a switch into the colour-naming task, but it is smaUer than the cost of a switch into the word-reading task. Meuter and AUport (1999) reported an analogous finding when subjects switched between digit naming in their dominant and nondominant languages. This paradoxical finding - larger reaction time costs for a switch into a better learned, more dominant task - is difficult to explain if switch costs reflect the duration of a process that reconfigures the cognitive system for the upcoming task. Why should a switch into an easier, better learned task take longer to complete than a switch into a less famUiar task? AUport et al. argue that the result can be explained if the primary determinant of the switch cost is the nature of the previous task. When subjects name the colour of an incongruent Stroop stimulus, AUport et al. hypothesise that inhibition of the word-reading task may be required. According to the TSI account, this inhibition wUl persist on a switch trial where word-reading is now appropriate, leading to a large switch cost. But in the absence of any requirement to suppress colour naming in order to perform the word-reading task, there wUl be no carryover of inhibition into colour naming switch trials, hence the smaU or absent switch costs.

2.6.2 Reverse Stroop interference

AUport et al. (1994, Experiment 5) also included a condition involving sequences of neutral stimuU (colour words shown in black for word reading, and "xxxxx"

times to incongruent and neutral stimuli, AUport et al. were able to assess the interference from the other task, both for switch and repeat trials. They found that, on nonswitch trials, there was a large interference effect in the colour- naming task (i.e. a large RT difference between incongruent and neutral trials), and Uttle interference in the word-reading task. This is the usual pattern of results in studies of interference effects in Stroop tasks (MacLeod, 1991). However, on switch trials, interference from colour naming was very much greater in the word-reading task, i.e. a "reverse Stroop effect" was now detectable. This provides evidence for an intrusion of the previous task (colour naming) into word-reading switch trials, supporting task carryover accounts of switch costs.

2.6.3 Effect of prior stimulus congruity

If switch costs reflect a carryover of task set from the previous task, it should be possible to modulate their size simply by changing the nature of the task Jrom

which the subject is switching. An experiment by Allport and Wylie (2000, Experiment 1; also reported in Wylie & Allport, 2000, Experiment 1) iUustrates the successful modulation of switch costs in this way. In this experiment, subjects again switched between word-reading and colour-naming tasks, and the main focus was the cost of a switch into the word-reading task. The experiment was carried out in three phases. First of all (phase one), subjects switched between word reading and colour naming, using neutral stimuli for both tasks. This gave rise to small, symmetrical switch costs in both tasks. Next (phase two), incongruent Stroop stimuli, instead of neutral stimuli, were used in the word- reading task. However, stimuli in the colour-naming task were kept identical (i.e., neutral). Despite the change from neutral to incongruent stimuli for word reading, switch costs were practically identical in these two phases. However, when subjects were also presented with incongruent stimuli for colour naming.

in phase three, the word-reading switch costs were increased by a factor of almost three. This was despite no change in the word-reading task itself. Thus, Allport and Wylie conclude that it is the task being switched from, and not the task being switched into, which is the primary determinant of the size of switch costs.

2.6A Longer lasting carryover effects

The evidence presented so far, supporting the TSI account of switch costs, could be explained by a short-term carryover effect, operating between immediately adjacent trials only. However, other studies have provided evidence for longer lasting carryover effects. For example, in an experiment by Mayr and Keele (2000; see also Mayr, 2002), there were three possible tasks. This permitted a comparison between two types of switch trial: a) a switch into a task different from the one performed on either of the two previous two trials (i.e. an "ABC' sequence) and b) a switch into the same task that was performed two trials previously (i.e. 'CBC'). In both cases, the final trial involves performance of task 'C' preceded by task 'B'. However, Mayr and Keele (2000) found that RT is slower on the final trial of a 'CBC' sequence than the final trial of an 'ABC' sequence. They claim that when subjects switch from one task to another, they inhibit the no-longer required task set. Thus, when they switch back into this task (i.e. in a 'CBC' sequence) they are slower to respond since the 'C' task set is still inhibited. Evidently, this phenomenon (referred to by Mayr and Keele, 2000 as 'backward inhibition') can survive at least one intervening trial.

An investigation into the longevity of carryover effects was carried out by Allport and Wylie (2000, Experiment 2; also reported in Wylie & Allport, 2000, Experiment 2). This experiment was based on the earlier finding, described above, that word-reading switch costs are larger when subjects switch from

colour naming with incongruent rather than neutral stimuli (Allport & Wylie, 2000, Experiment 1). Using the alternating runs paradigm, subjects switched between word-reading and colour-naming tasks. Each block consisted of a repeating sequence of 24 neutral trials, followed by 24 incongruent trials, followed by a further 24 neutral trials and so on. Thus, AUport and Wylie were able to examine to what extent the enhanced word-reading switch cost remained, or was dissipated, over the course of 24 trials with neutral stimuli for both tasks. The results indicated that the word-reading switch cost did indeed become smaUer over the course of each 24-trial mini-block of neutral trials, but nevertheless it was still enhanced at the end of this mini-block, in comparison with a similar experiment (AUport & Wylie, 2000, Experiment 1) where subjects were presented with neutral trials before being exposed to the incongruent condition. Thus, the carryover effect seems to be relatively long lasting (see AUport et al., 1994, Experiment 4 for further evidence of relatively long lasting carryover effects).

2.6.5 Task set inertia versus task set retrieval

In the TSI version of the task carryover account (AUport et al., 1994), the carryover effect was seen as a persisting activation or inhibition at the level of the task set. As such, it should not be affected by the identity of the stimuU presented in each task, since task sets, by definition, apply to aU task-relevant stimuli. However, AUport and WyUe (2000, Experiment 5) provided evidence for a strongly item-specific task carryover effect. Subjects alternated between short runs of colour naming and word reading, but only a subset of the stimuli that appeared m the word-reading task were also presented for colour naming. Thus, it was possible to compare two types of stimuU in the word-reading task: ^primed' stimuli, which had also appeared recently in the colour-naming task, and "unprimed' stimuU, which were only ever seen in the word-reading task.

Allport and Wylie found that the reaction time to primed stimuli on switch trials was slower than the reaction time to unprimed stimuli, but there was no reliable difference in reaction time between primed and unprimed stimuli on nonswitch trials. Thus, there was a greater switch cost when the stimulus on the switch trial was primed. This finding suggests that stimuh might themselves evoke task sets with which they were recently associated, even when this task set is not appropriate. As a result. Allport and WyUe (2000) have updated the TSI theory of Allport et al. (1994). They propose an alternative task carryover account, according to which stimuh are able to evoke recently associated task sets from memory (this will be referred to as the 'associative-TST theory). When a stimulus appears on one trial, associated with task A, and then reappears on a subsequent switch trial, requiring performance of task B, Allport and Wyhe hypothesise that task set A may nevertheless be evoked by the presentation of the stimulus (cf. perceptual "trigger conditions' in Norman & ShaUice, 1986). The resulting competition between task sets A and B may lead to an extended response selection process, explaining the enhanced switch cost obtained for word reading when the stimulus was recently presented for colour naming^.