In the target article, Juslin & Västfjäll (J&V) eloquently propose a fascinating theoretical framework and mechanisms to explain the emotional responses to music, but they do so predominantly from an adult perspective. I raise two issues: (1) In musical interactions, infants' emotional responses are elicited not only by the music they listen to but also by the information they perceive from several modalities, and these responses change with time. (2) In singing interactions, mothers convey the hierarchical structure of the song so that the infants create musical expectancy.

When mothers sing or play taped music in interactions, their infants not only listen to the music but are also exposed to information from multiple modalities. In a study on mother–infant interactions in musical contexts during the first year of life, Longhi (Reference Longhi2003) performed an in-depth analysis of the temporal structure of the songs mothers sing to their infants and on the partners' non-verbal behaviours. She found that when mothers sing and play taped music, they nod their head and bounce their body, as well as move and touch their infants' limbs and body. Thus, during musical interactions, mothers use several modalities (e.g., auditory, visual, tactile, and kinaesthetic), and these seem to affect the infants' emotional responses. For instance, when mothers were asked to sing to their infants without touching them, 3-month-old infants displayed mostly neutral emotional states, no engagement, frequent self-touch, and never smiled, compared to when the mothers sang touching them. Similar emotional responses were displayed by 4-month-old infants when mothers played taped music without touching them. Interestingly, the mothers' multimodal information seems to evolve with the infant's development. At 3 months, physical contact and face-to-face communication are crucial for the interaction, whereas, at 7 months, the focus of attention changes from the partners to the object, and toy activity, in particular. In this way, the mothers' multimodal participation is performed according to the infants' abilities.

Also, infants' emotional responses change during the first year of life. In fact, Longhi (Reference Longhi2003) found that, at 7 months, infants smiled longer and showed mostly happy and very happy emotional states compared to when they were 3 months of age. However, at 3 months, infants showed more engagement than at 7 months. Thus, the function of musical interactions and the emotions they induce seem to change in relation to an infant's age. Perhaps, at 3 months songs are central in the interaction, helping the partners to communicate. At 7 months, by contrast, songs are less crucial to the interaction, and they become a way of sharing and having a good time together. Therefore, musical interactions, either with live or recorded music, are a multi-sensory experience of different modalities where the infants' emotional responses are elicited by the music, as well as the integration of this information, and change with development.

An important mechanism featured in the theoretical framework is musical expectancy, which, contrary to what has been suggested, is observed early in life. When mothers sing to their 3-month-old infants, they emphasise the hierarchical structure of the songs. A detailed analysis of the temporal structure of the songs revealed that mothers emphasise the metrical and phrasing structure of the songs, both acoustically through their singing and behaviourally by synchronising with the beats relevant to the temporal structure of the song (Longhi, in press). Extending the duration of certain beats over others, the mothers mark the boundaries between phrases and within phrases, facilitating the infants' processing of the musical event. Through their singing, touching, and moving their own as well as their infants' bodies, the mothers use synchronised multimodal channels to provide a temporally coherent segmentation of the musical event (Longhi, in press). As the infants are sensitive to repetition and redundancy of information (Bahrick & Lickliter Reference Bahrick and Lickliter2000), this might be more effective for conveying the segmentation of the musical event so that the structure of the songs becomes predictable. In fact, at 3 months of age, infants seem to have a mental representation of the musical interaction, synchronising significantly more often with certain beats rather than others, in particular those in the middle and at the end of the phrases (Longhi, in press).

Another level of the temporal structure of the songs that has an impact on the interaction is the phrasing structure, which, together with musical tempo, seems to help infants to create musical expectancy. When analyzing the duration of the phrases of the song, Longhi (Reference Longhi2003) found that mothers modify the length of the phrase according to their position in the song across the different tempos. In particular, when using allegro tempo, mothers significantly extend the duration of the fourth phrase, whereas, with andante tempo songs, the second phrase is significantly longer. In this way, mothers make themselves more predictable to their infant. Moreover, musical tempos appear to play an important role in aiding the partners to organise and co-ordinate their responses with the beat, as well as with each other. So, during andante tempo, mothers and infants synchronise with each other very accurately and significantly match their responses in the third phrase of the song compared to the others. On the other hand, during allegro tempo, both partners perform numerous synchronous behaviours with the musical beat, but only the mothers synchronise significantly with it. These findings suggest that musical tempos, allegro and andante in particular, might be of importance in the interactions, and that infants might use these tempos to anticipate the structure of the musical interaction. In fact, mothers may more often sing songs at these tempos, because the actual pace of the songs enables both partners to temporally organise their behaviours and interlock with each other. Therefore, the hierarchical structure of the songs together with their tempos are vital in creating musical expectancy already when infants are 3 months of age – favouring the flow of the interaction between mothers and infants, and promoting their attunement and harmonious communication.

In sum, a developmental perspective, in particular regarding musical interactions, can help in further understanding the emotional responses to music and offer the new framework more supportive avenues for consideration and investigation than the ones suggested by Juslin & Västfjäll.