Procedural memory; Insights to resolve the importance of time

So far, the behavioural and the neurophysiological evidence supporting the role of sleep in the offline processing of declarative memory has been summarized, and it is clear that the behavioural outcome of this consolidation process is reduced forgetting over the retention interval and as a result, better maintenance of recall performance. However, recent studies using procedural tasks have suggested that sleep is required not simply for maintaining performance, but is necessary for memory enhancement and behavioural improvement (see Walker & Stickgold, 2006 for a recent review).

Walker and colleagues (2002) trained their participants on a sequential finger- tapping task at either 10:00 or 22:00, and their performance was retested 12 hours as well as 24 hours later. After the first retention interval when the morning leaming group stayed awake, their performance was similar to that observed at the end of the leaming session, but after a noctumal sleep in the second retention interval, they showed marked improvement in performance. This ovemight improvement was also observed in the evening leaming group in the first retest, and their performance did not improve further after they stayed awake in the following 12 hours. Hence, newly acquired procedural memory was enhanced during sleep, but not during wakefulness. Memory enhancement is not limited to the first night after leaming but continues at least for three nights (Walker et al., 2003), and is supported by ovemight changes in the involvement of various brain stmctures which together produce more accurate and faster motor performance and reducing the need of conscious spatial monitoring (Walker, Stickgold, Alsop, Gaab, & Schlaug, 2005).

In another interesting study by the same research group (Walker, Brakefield, Hobson, & Stickgold, 2003), participants learned two motor sequences immediately after another. Both the accuracy and speed in typing the second sequence improved across the 24-hour retention interval, whereas for the first sequence, there was no apparent increase in accuracy, indicating that sleep-dependent enhancement of the first sequence was disrupted and interfered with by the acquisition of the second sequence. However, if leaming of the second sequence was delayed for 6 hours (participants stayed awake in the interim), ovemight improvement was found in both- sequences. These researchers concluded that newly acquired procedural memory has stabilized and become resistant to interference within the first 6 hours after leaming even during wakefulness, but memory enhancement was only facilitated during sleep.

Based on these findings and after summarizing existing ones in a detailed literature review. Walker (2005) proposed a two-process model in which consolidation of procedural memory involved both a time-dependent process of stabilization and a state-dependent (more specifically, sleep-dependent) process of enhancement. Newly acquired procedural memory is stabilized over time, regardless of physiological state, resulting in performance maintenance, while only during sleep can procedural memory be enhanced, leading to performance improvement over the retention interval even without further practice.

Can Walker’s (2005) model help explain findings regarding sleep and declarative memory consolidation? From 1885 when the forgetting curve of verbal materials was first published (Ebbinghaus, 1885) until the 1990s, improvement in declarative

performance over a retention interval has been rarely documented. Even when the retention interval involved sleep, as summarized previously, declarative memory performance either was maintained or deteriorated more slowly than across an equivalent period of wakefulness. In Walker’s (2005) terms, the newly acquired declarative memory is stabilized during sleep.

However, improvement^ in declarative performance across noctumal sleep was first reported in 1997 by Plihal and Bom. As mentioned previously, the two researchers contrasted the effects of sleep in the first and second halves of the night on paired- associate performance similar to Ekstrand and colleagues’s studies (Barrett et al., 1972; Fowler et al., 1973; Yaroush et al., 1971). But instead of using a daytime wakefulness control group, Plihal and Bom used two better wake control conditions which were conducted at the same time of day, i.e. over the first and the second halves of the night, as the respective experimental sleep conditions (see Figure 1.2 - bottom right panel for study design). They reported an impressive 32.4% increase in the number of correctly recalled word pairs over a 3-hour sleep episode in the first half of the night, while only a 16.5% improvement was found in the corresponding wake group. The effect of sleep in the second half of the night was also relatively small (11.1%) and did not differ from that observed in the corresponding wake group (12.2%). Their PSG data showed that the amount of SWS was five times more in the first than in the second half of the night, so it was concluded that declarative memory enhancement was facilitated during SWS. Since then, there has been a rapidly

^ The question about whether the improvement in declarative performance across noctumal sleep and

daytime nap documented in the existing literature indeed reflects memory enhancement will be further discussed in section 1.4.1.

increasing number of studies that used the paired-associate task and replicated this enhancing effect of sleep (e.g. Gais & Bom, 2004; Marshall et al., 2006; Tucker et al., 2006; Wilhelm, Diekelmann, & Bom, 2008).

This difference between earlier and more recent studies about the effects of sleep on declarative memory consolidation is striking since they all used the same task, i.e. the paired-associate task. Stickgold (2004) attributed this dramatic change in empirical findings to the differences in the leaming materials used. He pointed out that behavioural improvement was observed across sleep only in those studies that used semantically related word pairs, the associations of which were pre-existing and did not require the mediation of the hippocampus. In contrast, recall performance did not improve over sleep when semantically unrelated word pairs were used because hippocampal mediation was required for these newly formed associations. Using Walker’s (2005) terms, sleep facilitates the stabilization of recently acquired, hippocampally-dependent declarative associations and the enhancement of pre­ existing, hippocampally-independent declarative associations.