Medición de presión

Introduction

Experiment 3 investigated the conditions under which ERPs associated with old items in an exclusion task exhibit greater negativity than those associated with truly new items over mid and right parietal sites from 800 msec until the end o f the recording epoch. The fact that this effect was not detected in Experiment 1, but was observed in Experiment 2 for preexposed correct rejections relative to non-preexposed correct rejections suggests that some kind o f additional process was recruited during test performance in Experiment 2 that was not recruited in Experiment 1. It is important to establish what the nature o f this process may be in order to understand the dynamics o f recognition memory more fully. Behavioural data indicated that discrimination o f studied items from unstudied lures was poorer in Experiment 2 than in Experiment 1. It was therefore hypothesised that the late negative wave over mid and right parietal sites may be elicited when discrimination o f source information is relatively difficult. However, it was not possible to determine whether this effect occurred because greater positivity was elicited by new items in Experiment 2 than in Experiment 1 (i.e. a novelty effect) or because greater negativity was elicited by old items in Experiment 2 than in Experiment 1.

Negativity over mid and right parietal sites has been observed previously if infrequently. W ilding and Rugg (1996) recorded ERPs during an exclusion recognition memory task, and reported greater negativity associated with both included and excluded old items relative to new items over mid and right parietal sites, onsetting at approximately 800 msec. This effect was larger in amplitude for correctly excluded old items (‘nontargets’) than for included items (‘targets’). It was proposed that this effect may be response-related as opposed to mnemonic, as the amplitude o f the late negative wave was positively correlated with RT (Wilding & Rugg, 1996). Comparable findings from other studies requiring the retrieval o f source information (Rugg, Schloerscheidt, Doyle, Cox, & Patching, 1996; Nessler,

Mecklinger, & Penney, 2001) supported this hypothesis (see Chapter 3). However, a further study reported right parieto-occipital negativity for correctly identified old items relative to new items in an exclusion task, despite the fact that RTs to these two classes o f item did not differ significantly (Cycowicz, Friedman, & Snodgrass, 2001). The finding that this

negativity was not elicited by old items in an inclusion task led to the argument that this effect is related to the requirement to differentially respond to old items depending on their source, a task that usually requires additional monitoring and evaluation processes, and that the late negative wave therefore reflects search or evaluation processes associated with the recollection o f contextual details (Cycowicz, Friedman, & Snodgrass, 2001). Such an account may explain why items eliciting this negativity are often associated with longer RTs. However, late parietal negativity is not always elicited by items for which source information is retrieved, whereas it is occasionally elicited when subjects are not required to retrieve source information at all (Donaldson & Rugg, 1998). It therefore appears that the late negative wave is not a direct correlate either o f long RTs, or o f the retrieval o f source information, although it may reflect processes indirectly related to both o f these factors.

Two other possible explanations present themselves. It may be the case that the late negative wave reflects some form o f response conflict that occurs when items are recognised either on the basis o f familiarity or noncriterial recollection, but when this non-diagnostic retrieval does not allow subjects to respond until diagnostic information is recollected. This effect may therefore be elicited when an item is recognised as being old, but when an ‘old’ response must be suppressed until contextual details from the prior presentation have been retrieved. According to this hypothesis, although the process reflected by the late negative wave may be highly correlated both with RT and with the successful retrieval o f contextual details, it is not causally linked with either o f these factors. This hypothesis is consistent with the finding that the late negative wave is associated with activations in the anterior cingulate cortex, a finding which led to the argument that both effects may reflect response competition in situations in which the likelihood o f response error is high (Mecklinger, 2000). Similarly, the finding that this effect was only elicited in Experiment 2, when discrimination was poorer than in

Experiment 1, is consistent with this interpretation. This hypothesis does, however, rest on the assumption that the effect observed over parietal sites is in fact a negative wave elicited by old items, rather than a positive wave elicited by new items. If the effect is in fact a positive wave elicited by new items, it may be the case that this effect is actually an enhanced response to the subjective novelty o f an item. According to this hypothesis, as discrimination o f old items from various sources becomes increasingly difficult, truly new items are

perceived to ‘pop out’ on the basis o f their novelty. I f this is the case, one would predict that these items would be associated with faster RTs, a finding that has been reported by a number o f the studies described above.

Experiment 3 attempted to delineate the functional significance o f the late negative wave in order to understand the contribution o f the process/es reflected by this effect to performance during an exclusion task. The hypothesis that this effect is elicited only when discrimination o f source is relatively difficult was directly tested in a between subjects design. ERPs were recorded while each group completed an exclusion task similar in design to that employed in Experiments 1 and 2. One group performed a shallow encoding task in study phase 1 ( ‘shallow’ group), while the second group performed a deep encoding task (‘deep’ group). H alf o f these words were repeated in study phase 2. At test, both groups were required to respond ‘old’ to items from study phase 2 (‘targets’), and to exclude all other items. The between group manipulation o f levels-of-processing during study phase 1 was intended to vary the ease with which targets could be discriminated from nontargets, as it was anticipated on the basis o f the findings from Experiments 1 and 2 that study phase 1 items should cause more interference when deeply encoded than when shallowly encoded. Therefore, it was predicted that a late negative wave would be observed for old items in the deep group, but not for old items in the shallow group. This design also permitted an evaluation o f whether this effect is elicited by new words or by old words, as these items could be directly contrasted between the two groups.

Also o f interest was whether the parietal old/new effect would be elicited by all correctly classified old items or not. According to Jacoby (1991), both correctly identified targets and correctly identified nontargets should be recollected, as correctly excluded nontargets are assumed to be excluded on the basis o f recollection. Consistent with this assumption, a number o f ERP studies employing exclusion tasks have reported significant parietal old/new effects both for correctly endorsed targets and for correctly excluded nontargets (Rugg, Brovedani, & Doyle, 1992; W ilding & Rugg, 1997; Cycowicz, Friedman, & Snodgrass, 2001). In some cases these effects have been smaller for nontargets (Rugg, Brovedani, & Doyle, 1992; W ilding & Rugg, 1997), whereas in others the parietal effect elicited by targets and nontargets have been equivalent in amplitude (Cycowicz, Friedman, & Snodgrass, 2001). In the light o f these studies, it was somewhat surprising to observe that deeply studied preexposed correct rejections did not elicit a parietal old/new effect in Experiment 2, as one would have predicted that these items would only be correctly classified if their source was correctly recollected. This finding was not without precedent; a previous study had reported that correctly excluded nontargets in a continuous recognition memory paradigm did not

elicit a parietal old/new effect when discrimination was good (Dywan, Segalowitz, &

Webster, 1998). It was therefore o f great interest as to whether correctly excluded nontargets in Experiment 3 would elicit a parietal old/new effect or not.

Similarly, it was o f interest which responses would elicit the putative correlate o f familiarity; the early frontal effect. As familiarity is thought to be an automatic and acontextual form o f recognition, one may predict that the ERP correlate o f familiarity should be elicited by old items regardless o f their source or response requirement, and that this effect should therefore be observed for both targets and nontargets. However, no early frontal effect was detected for preexposed correct rejections in Experiments 1 and 2, although an early frontal effect was observed for preexposed false alarms. It was argued in Chapters 5 and 6 that only those preexposed unstudied items whose familiarity level was above response criterion (and which were therefore misattributed to the target study phase) elicited an early frontal effect, and that correctly identified preexposed unstudied items were associated with lower levels o f

familiarity. If subjects were to adopt the same strategy in Experiment 3, an early frontal effect may not be elicited by nontargets.

Method

Subjects

Each group consisted o f 18 subjects, resulting in a total o f 36 subjects. 44 subjects

participated, and 8 sets o f data were discarded prior to analysis as they failed to provide more than 16 artefact-free trials for one or more o f the critical ERP conditions. The 18 subjects who contributed data to the ‘deep’ group consisted o f 11 females and 9 males, as did the 18 subjects contributing data to the ‘shallow’ group. Subjects were alternately allocated to each group. The average age o f the deep group was 20 years and ranged from 1 8 - 2 5 years, whereas the average age o f the shallow group was 21 years and ranged from 1 8 - 2 7 years. These two groups did not differ significantly in age.

Stimuli and Design

Experiment 3 employed a mixed 3 x 2 design, in which the factor o f group was crossed with the within subjects factor o f item type (i.e. target/nontarget/new), and consisted o f three phases; study phase 1, study phase 2 and test. Figure 7.1 shows the design o f Experiment 3.

The term ‘targets’ refers to old items from study phase 2, ‘nontargets’ refers to old items from study phase 1 alone, and ‘new ’ refers to unstudied items. This nomenclature is adopted throughout the remainder o f this thesis, and replaces the terminology employed in Chapters 5 and 6 because the factor o f preexposure was no longer fully crossed with the factor o f studied/unstudied. Non-preexposed studied items were not employed in Experiment 3 because these were not required for the principal aim o f this experiment (i.e. delineation o f the functional significance o f the late negative wave).

Time

STUDY PHASE 1

STUDY PHASE 2