Participants listened to mechanical (precisely timed) and human performed (with natural timing variability) versions of a piece of rhythmic music, Clapping Music, and their EEG was recorded to measure neural entrainment (the delta-band ITPC) to the 12 unique rhythms of the piece. A separate group of participants rated each of the rhythms, from mechanical and performed versions, and provided their subjective evaluations of rhythms on complexity, pleasure, beat, and groove. Neural entrainment, as measured by the delta-band ITPC, was greater when the rhythms were presented in a temporally precise, mechanical version compared to a performed version. We suggest that this difference in neural entrainment is likely due to the increased temporal precision of the mechanical rhythms—because the stimulus is more consistently timed, the entrained oscillations are more consistent, resulting in greater ITPC. However, overall subjective ratings of complexity, groove (the desire to move along to the rhythms), beat perception, and pleasure did not differ between performed and mechanical versions. Importantly, we observed relationships between neural entrainment and perceived complexity and groove only for the performed rhythms, but not for the precisely timed ones. Although subjective groove and complexity ratings did not differ between mechanical and performed rhythms, it may be that the functional relationships between neural entrainment and the experiences of subjective groove and complexity are different for the two types of rhythms. For example, it may be that the temporal variability of performed rhythms requires greater use of neural entrainment to attend to and assess the rhythms, or that when rhythms are perceptibly human-generated, subjective perception influences neural entrainment in a top–down fashion, possibly mediated by attention (i.e., a rhythm may be more salient when it is evidently produced by humans than if it is computer generated). It is further possible that the presence of relationships between neural entrainment and subjective perception for one but not the other stimulus type is due to different underlying influences on neural entrainment. This account is bolstered by the fact that both stimulus regularity (stronger for mechanical rhythms) and attention (possibly stronger for performed rhythms) can both increase neural entrainment (Fujioka et al. 2012; Lakatos et al. 2008; Calderone et al. 2014), but of those two factors, attention is more plausibly related to subjective perception. Alternatively, relationships between subjective perception of rhythms and neural entrainment to those rhythms could be present for mechanical rhythms but unobserved, because the relationship is so subtle that it is dominated by the relatively strong stimulus-driven component of neural entrainment. The present data cannot distinguish between the likelihood of these accounts (or others), so further research is needed to clarify the relationships between neural entrainment to stimulus characteristics and subjective perception of musical rhythms.
Groove, complexity (e.g., nPVI, syncopation), and pleasure have been shown to correlate with each other (Witek et al. 2014), and our behavioural ratings showed a similar pattern. Perceived complexity was correlated with groove for both mechanical and performed rhythms, while pleasure was correlated with complexity and beat only for performed rhythms. Pleasure and groove did not correlate for either version. The limited observed relationships between rated pleasure and groove, complexity, and neural entrainment may be due to the limited stimulus set (i.e., all short rhythms performed by clapping). The enjoyment of these particular rhythms was expected to be lower, and less variable, compared to the popular music recordings or drum kit performances used in the previous studies of groove (Janata et al. 2012; Witek et al. 2014). Perceived complexity (ratings) and objective complexity (nPVI) were only correlated for performed, and not for mechanical, rhythms, despite complexity ratings not differing between rhythm conditions. However, it is not clear why the relationship between objective rhythmic complexity (nPVI) and perceived complexity would be different for performed vs. mechanical rhythms.
Surprisingly, ratings did not differ between performed and mechanical rhythms. We expected that the expressive timing in performed rhythms would lead to higher ratings, particularly of pleasure and groove, based on the previous literature (Hellmer and Madison 2015; Hennig et al. 2011; Räsänen et al. 2015). It may be that the particular performed and mechanical stimuli used here are not sufficiently distinct in temporal variability to elicit differences in explicit ratings despite their different association with neural entrainment. This is consistent with other EEG research, showing that electrophysiological measures are, in some cases, more sensitive than behavioural measures (e.g., Francois and Schön 2011; Peretz et al. 2009). On the other hand, pleasure was significantly correlated with complexity and beat only in the expressive condition, suggesting that expressive timing may be a necessary condition for complexity and meter perception to exert influence on pleasure.
Neural entrainment was greater for rhythms with greater objective rhythmic complexity (nPVI). The three rhythms with lowest objective complexity (nPVI = 0; rhythms 2, 7, and 12) elicited the lowest levels of neural entrainment. Although this may seem contrary to the expected positive relationship between strict regularity in a stimulus and neural entrainment, it is worth noting again that we measured neural entrainment only in the delta band (1–4 Hz), a band that excludes the stimulus rate in these three isochronous rhythms (5.33 Hz).
It is worth considering our findings in the context of recent interest in entrained neural oscillations, its disentanglement from regularly occurring evoked neural responses, and its possible functions [see recent reviews by Zoefel et al. (2018), and Haegens and Golumbic (2018)]. We showed that phase locking in the delta-band EEG to rhythms depends both on objective properties of the rhythms and on subjective perception of them. We suggest that these results do not reflect differences in evoked responses arising after sound onsets in rhythms. Although ITPC was greater for the more precisely timed mechanical rhythms than for performed rhythms (a difference which could be due—at least in part—to sound-evoked responses that were more precisely regular), ITPC was not greater for the more structurally regular (less complex) rhythms, and, in fact, was greater for more complex rhythms when they were of the performed type. This suggests that differences in endogenous neural oscillations could contribute to the observed differences in ITPC.
Endogenous neural oscillations are thought to support temporal predictions (Lakatos et al. 2008; Arnal and Giraud 2012; Calderone et al. 2014; Zoefel et al. 2018), and can be driven by temporal predictability in stimulus streams. Here, we showed that neural entrainment was greater for more temporally regular (mechanical) rhythms, but also for more complex rhythms (both objectively and subjectively complex) if they were performed rather than mechanical. The correlations between subjective experience and neural entrainment to performed rhythms may support a recent proposal that complexity in musical rhythms is associated with bodily movement and pleasure by way of predictive neural mechanisms, and that beat-entrained movements not only elicit pleasure but aid sensory predictions (Vuust and Witek 2014). Thus, it may be that the predictive function of entrained endogenous oscillations supports neural and cognitive processing of complex rhythms, leading to the correlations between entrainment to performed rhythms and their perceived groove, and both objective and subjective complexity.
Subtle timing variation in rhythms, or micro-timing, may be related to the observed differences (and absence of differences) for performed and mechanical rhythms which differ in terms of temporal variability. The previous work has considered whether or not micro-timing in musical rhythms is related to groove, with mixed results (Butterfield 2010; Davies et al. 2013; Kilchenmann and Senn 2015). When micro-timing effects on groove have been shown, they tend to be related to systematic timing variation rather than ongoing variability arising unintentionally from natural human performance, and in relation to specific music genres. The lack of any difference in groove between the performed and mechanical conditions may be related to the fact that the stimuli are from a piece of music in a style not usually associated with groove (minimalist twentieth century art music), in the way that jazz and funk are, for example. Micro-timing may, however, be related to the differences between performed and mechanical rhythms in terms of relationships between neural entrainment and subjective perception of groove: micro-timing may impact attention, providing a functional link between neural entrainment and subjective perception, as discussed above.
Of note, the presentation order of rhythms was constant across EEG participants (the rhythms were presented as a musical composition to keep the ecological validity of musical listening), but differed across participants completing the behavioural experiment (randomized order of individual rhythms). While we do not expect that the order of rhythm presentation casts significant doubt on our main conclusions (as the order was the same for both mechanical and performed rhythms in the EEG experiment, and did not systematically differ between conditions in the behavioural experiment), a previous study showed that the rhythms (or rhythmic figures) of Clapping Music are more easily differentiated (rated as less similar) when heard within the context of the entire piece of music rather than as isolated pairs (Cameron et al. 2017). Therefore, it is possible that relationships between perceptual ratings and neural entrainment would be stronger if presentation order was the same in the two experiments (although this would either reduce ecological validity in the EEG experiment or risk introducing order effects on subjective ratings).
While we believe that the use of real music provides ecological validity to the study of rhythm perception and elicits greater engagement from participants, the need certainly exists to use a broader range of stimuli to investigate in further breadth and detail the relationships between neural entrainment to, and perception of, musical rhythms. For example, using a larger set of rhythms with a range of complexity (nPVI) that reached higher levels of complexity might replicate the correlation between complexity and neural entrainment observed here, but might, instead, reveal an inverted-U relationship between neural entrainment with complexity, as observed previously for groove and preference (Witek et al. 2014).
All participants were trained musicians in our two experiments; both behavioural and neural differences associated with rhythm and beat perception have been found between musicians and non-musicians (e.g., Drake et al. 2000; Grahn and Rowe 2009), including an enhancing effect of musical training on neural entrainment to music and rhythms (Doelling and Poeppel 2015; Stupacher et al. 2017). In addition, participants were primarily trained in Western music and were all living in the UK (i.e., as a sample they did not represent global cultural diversity), and learning, enculturation, and experience are known to influence both musical rhythm perception (Hannon and Trehub 2005; Hannon and Trainor 2007; Hannon et al. 2012; Stevens 2012; Cameron et al. 2015; Bouwer et al. 2018; Polak et al. 2018) and neural entrainment to music, rhythms, and speech (Doelling and Poeppel 2015; Stupacher et al. 2017; Song and Iverson 2018). Therefore, although the phenomena of interest (beat perception and tendency to entrain to musical rhythms) are found widely across the world, in all cultures, and do not require training, enculturation and training may influence, and thus limit the generalizability of, the observed relationships between the perception of and neural entrainment to musical rhythms.
Altogether, we demonstrate links between the subjective experience of, neural entrainment to, and complexity of performed (but not mechanical) musical rhythms. The causal links between these factors and measures remain to be understood—for example, neural entrainment could either cause or arise from the desire to move while listening to rhythms—but the presented results contribute to understanding the seeming magic that music exerts on our senses, bodies, brains, and lives.