4.2 P2 Adaptation: A Model Adjustment Effect
The P2 amplitude displayed a continuous increase over the first ten tones, contrary to our hypothesis and findings from previous studies where a steep decrease followed by a plateau was observed (Rosburg & Sörös, 2016; Rosburg et al., 2010). However, this continuous increase could be interpreted as repetition positivity (RP), as seen in studies by Costa-Faidella et al. (2011a, 2011b). Additionally, Recasens et al. (2015) reported repetition enhancement in a time window close to P2 (230–270 ms after stimulus onset).
Discrepancies in these findings may stem from differences in paradigms, such as variations in participant expectations and ISI. For instance, in studies where participants could easily predict the tone pattern (e.g., Rosburg & Sörös, 2016), a sharp decline followed by a plateau in P2 was observed. Conversely, when tone predictability was low, as seen in the present study and others (Costa-Faidella et al., 2011a, 2011b; Recasens et al., 2015), a continuous increase was noted. Hence, expectations may play a role in modulating the adaptation pattern revealed by P2. Moreover, differences in the time allowed for recovery from adaptation may contribute to the discrepancies. Shorter ISIs used in previous studies (e.g., Costa-Faidella et al., 2011a, 2011b; Recasens et al., 2015) may lead to continuous P2 increase, while longer ISIs (e.g., Rosburg & Sörös, 2016; Rosburg et al., 2010) may result in a sharp decrease followed by a plateau.
Furthermore, variations in experimental paradigms used across studies can also contribute to the inconsistencies. Studies by Costa-Faidella et al. (2011a, 2011b) and Recasens et al. (2015), like the present one, employed a roving paradigm, potentially enhancing memory trace due to unpredictable tones. In contrast, paradigms with discontinuous stimulus sequences or predictable standard tones, such as the stimulus pair or traditional oddball paradigms, may require less memory trace strengthening (Cooper et al., 2013). This underscores the link between the RP and memory trace, particularly in paradigms like roving where sensory memory must be constantly updated (Cooper et al., 2013). As these postulations lack direct comparison, factors influencing the relationship between RP and adaptation warrant further investigation. Notably, the adaptation effect observed in the stimulus pair and traditional oddball paradigms may not generalize to roving paradigms or situations where expectations are controlled.
As previously discussed, the increase in P2 can be interpreted as an RP. Despite the negative frontal-central component observed in the topography of the P2 subsequent adaptation (Figure 6b), this was due to the reverse calculation method of the adaptation effect (2nd tone – final tone instead of final tone – 2nd tone). Nonetheless, the RP found in the present study exhibited a delayed onset compared to a previous study by Haenschel et al. (2005), wherein they compared RP across two, six, and 36 standard repetitions in healthy adults. Their findings indicated that RP, occurring roughly from 50 to 250 ms post-stimulus, was larger in conditions with greater standard repetitions, with RP contributing predominantly to the MMN. Hence, they suggested that RP serves as an ERP correlate of adaptation, a mechanism that facilitates memory trace formation in the A1.
However, it remained uncertain whether the heightened RP and MMN observed after 36 standard repetitions in Haenschel et al. (2005) were attributable to an extended memory trace or an enhanced precision in participants’ expectation. After a certain duration, participants might readily expect the appearance of a deviant following 36 standard repetitions if none had occurred at the 2nd and 6th positions. In contrast, in the present study, where participants could not predict the occurrence of deviants, RP, characterized by a continuous amplitude increase across the stimuli, was still evident. However, it manifested with a later onset within the P2 time window, compared to the roughly 50 ms onset observed in the study by Haenschel et al. (2005).
While our findings do not challenge the interpretation made by Haenschel et al. (2005) that RP links to adaptation and memory trace formation, the present findings suggest that the latency of RP may be influenced by the precision of the prediction. Specifically, a delayed onset may manifest when prediction precision diminishes. Conversely, an earlier onset of RP may occur due to a heightened precision in the prediction model, as in Haenschel et al. (2005). However, this postulation remains speculative and warrants future research, given the parameter discrepancies between the two studies, as discussed below.
The differences in experimental parameters between the two studies, including intensity, pitch, and inter-stimulus interval (ISI), might lead to the weaker N1 and P2 components found in the current study. Admittedly, while the P2 amplitudes in our study exhibited a positive-going trend, they did not reach positive values as observed in Haenschel et al. (2005), where P2 exhibited stronger amplitudes. One plausible explanation is the difference in stimulus intensity; Haenschel et al. used stimuli with a higher intensity (80 dB) compared to our study (70 dB). Past research indicates that increased stimulus intensity often leads to larger N1 and P2 amplitudes (Adler & Adler, 1989; see Crowley & Colrain, 2004, for a review). Moreover, Haenschel et al. (2005) used more high-frequency stimuli with a boarder range on average (100 to 5,000 Hz), compared to our study’s lower frequency and narrower range (500 to 800 Hz). Studies have shown that as frequency increases, N1 and P2 amplitudes decrease (Wunderlich & Cone-Wesson, 2001), which could explain the weaker N1 and P2 observed in our study.
Regarding inter-stimulus interval (ISI), although it is known to positively affect N1 and P2 amplitudes, it is unlikely to be the primary cause of the discrepancy between our study and Haenschel et al. (2005). Our study used a longer ISI (580 ms) compared to Haenschel et al. (2005) (300 ms). In addition, they employed a between-train interval of 500 ms, whereas we played the sounds continuously. Although the ISI was longer in the present study, our study did not exhibit larger N1 and P2 amplitudes, as one might expect if ISI was a significant factor. More importantly, the disparity in ISI between the two studies was minimal. Previous research has indicated that for each 10-fold rise in ISI, N1 or P2 amplitudes increase by approximately 5.6 µV (Crowley & Colrain, 2004). Hence, the weaker amplitudes observed in our study are mainly attributed to differences in stimulus intensity and pitch.