Photo of Manuela Filippa taken by Craig Cutler. This issue of Acta Paediatrica presents a remarkable paper by Lejeune et al. on the negative effects of the recorded female voice on the tactile memory and discrimination of preterm infants with a postconceptional age of 28–35 weeks 1. During their stay in the neonatal intensive care unit (NICU), the infants had to discriminate between a familiar and unfamiliar object while 26 were simultaneously exposed to silence and two groups of 24 were exposed to a recorded female voice played at two different volumes, +5 and +15 decibels, inside their incubator. The woman was asked to speak in a pleasant and age-appropriate way, but the content was unrelated to the handling task. The major findings were that being exposed to the recorded female voice had a negative impact on their tactile sensory learning and these negative effects were not linked to the volume of the recording. These findings raise important questions about early sensory stimulations in the NICU. We are now seeing more, high-quality intervention projects that explore preterm infants’ sensory stimulation and these are largely based on the theoretical framework of the environmental enrichment. Having said that, there are still limitations with these studies, including the lack of consistent outcome measures 2. However, in the light of growing evidence, especially in animal models, the manipulation of early sensory experiences can demonstrate both beneficial and detrimental effects. The outcomes depend on specific conditions: when they take place, what type of sensory stimulation is provided, how is it administered, how long it lasts and the individual reactions of the infants. The relative immaturity and the specific limitations of the development of each of the infants’ senses might have important, positive and adaptive functions in subsequent phases of their perceptual and behavioural development. The natural stimulation of a particular sensory system, at a precise time, can sustain its development. However, if the same stimulus is repeated or occurs during a different phrase in the infants’ development, it can have detrimental effects on the maturation processes of that system. Hyperdevelopment of certain functions can also have detrimental effects on proximate functions that are not equally stimulated. Neonatal intensive care unit interventions based on auditory stimulation have increased dramatically and these have included both live and recorded music and mothers’ voices 3. Trained professionals, including music therapists, have developed a specific interest in using sound in NICUs 4. Despite this increased interest, and greater data from published results, a number of questions remain unanswered. These include which recommendations about auditory stimulation for preterm infants can be put into practice by healthcare practitioners. We also need to know the limits of these interventions, the most appropriate way to intervene and who should intervene 2. The paper by Lejeune et al. calls for us to reflect on the limits of a direct application of the environmental enrichment theory on early interventions in the NICU 5, which include the assumption that enriching is always an improvement. However, this paper 1 and a critical review 6 have pointed out the risks of increased early auditory exposure on preterm infants’ sensory experiences. The NICU sensory environment lacks meaningful and contingent sensory inputs. It lacks the significant presence of the live mother's voice, which decreases dramatically after preterm birth when compared to the uterine environment. It also generally lacks the early and continuous multimodal contact with parents, as humans are one of the very few mammals where offspring can survive prematurity even when they are separated from the primary caregiver. More specifically, the Lejeune et al.'s paper prompts us to consider the types and characteristics of additional auditory sensory stimulation in the NICU. In this loud acoustic environment, reported by the authors as exceeding the pressure levels recommended by the American Association of Pediatrics, specific recommendations should be taken into account. The additional stimulation of a recorded voice should evaluate the possible reverberation phenomena that could occur in the incubator 7. Moreover, it is suggested that a minimal signal-to-noise ratio threshold of 5–10 dBA is required to allow the infants to detect the stimulus 8. The main finding of this paper was that preterm infants could only effectively discriminate between two objects by handling them in silence and that provided a better learning condition than an unconnected female voice. The distinction between a vocal stimulus that is related, or unrelated, to the tactile task being performed by the infant and that is modulated on his reactions could play a crucial role in how they perform. In the present study, the phrases repeated by the female recorded voice bore no relation to the task the baby was expected to perform and it was not connected to the infant's reactions. In normal conditions, when a parent interacts with their preterm infant in the incubator, they use gestures and voices simultaneously in a related way. For example, they intuitively soothe their babies with a descending vocal profile, which diminishes in pitch or intensity, and combine this with a descending gesture, caressing the preterm infant's face with a finger from the top of the head to the neck. In the same way, they may imitate the infant's gestures with their voice. The simultaneous use of two or more sensory modes to convey the same message is perfectly tuned to the preterm infant's perceptual abilities, which are multimodal in early development. In fact, during early periods of brain development, there are redundant connections, for example, between the cochlea and the visual cortex and the retina and the auditory cortex 9. These links diminish during development, but multimodal perception still remains active during infancy. The research that has been performed so far, including the current study, shows that constructing an appropriate sensory environment in the intensive care unit should be guided by a bioecological model 10. Preterm infants are not isolated in their learning experience during this phase of their development. They are sustained by shared multisensory experiences that guide their own personal development and these are characterised by precise rhythms, tempos, silences and linguistic elements. Part of this process is for infants to screen out what their sensory systems do not need to process at a particular point in time, as this is a fundamental part of how they understand their environment. They can then selectively attend to the sensory tasks they need to. As the authors of the current study state in their discussion, the female voice that was unrelated to the task and unrelated to the infant's reactions had a detrimental effect on the tactile learning process and impaired the preterm infants when they needed to discriminate between the features of the two shapes they were presented with. Future research is needed so that we can understand whether live parental voices, directed and in careful interaction with the preterm infant, can facilitate this recognition task and, in general, can impact on preterm infants’ cognitive and language development. The author has no conflict of interest to declare.