Selective attention involves attending to task-relevant information and inhibiting task-irrelevant information. While spatial priming is known to efficiently shape selective attention, the nature of language-mediated effects on selective attention is not well-understood, particularly in young toddlers. We com- pare the impact of language-mediated and spatially-mediated attention in an eye-tracking paradigm in which two objects are presented in one of four possible locations and one of the objects is highlighted. The impact of labelling on attention orienting during a prime phase was tested in a subsequent probe phase, where either the identity, location or both were manipulated, and compared to the impact of spatial priming. To elucidate the role of development on these effects, the study was conducted with 18- and 26-month-old toddlers. The results revealed that both language-mediated and spatially-mediated priming lead to attention orienting during the probe phase: at- tended information during the prime phase facilitates attention during the probe phase while ignored features are inhibited. However, in contrast to spatially-mediated attention during the prime phase, language-mediated attention can override these inhibitory effects. The impact of language on overcoming inhibitory effects is particularly noteworthy in the older age group.

Most of the recent success in machine learning has been achieved in supervised learning and predicated on the availability of large amounts of labelled training data. On the other hand, effectively using readily available unlabelled data has proven a much more difficult endeavour. In contrast to algorithms, infants spontaneously learn from the available sensory information without explicit instructions, supervision or feedback. Thus, infant learning can be viewed as a highly successful approach to unsupervised learning.  In this work, we explore the parallels between infant learning and recent successes of unsupervised machine learning in the area of contrastive learning. We examine how the principles of infant learning and developmental cognitive neuroscience can inform and inspire the development of novel contrastive learning algorithms. We focus on the phenomenon of category learning and explore how these principles can be applied to better understand and improve contrastive methods.

During the first year of life, infants develop a remarkable ability to group objects based on their similarities and differences. This ability of category formation represents one of the main mechanisms underlying the organisation of the semantic system. Early categories are formed spontaneously, in a non-supervised fashion and this type of category acquisition remains present even when more sophisticated forms of supervised category learning emerge. Even though there are various models of categorisation mechanisms across the lifespan, there is a gap in the research investigating implicit categorisation at different stages of cognitive development. Therefore, the aim of the current study was to compare processes of spontaneous concept formation in infants and adults using an experimental paradigm based on novelty preference. We discovered that both infants and adults show evidence of category learning (Experiment 1), though with different amount of training being needed to achieve the task. Adults successfully categorised objects already after a single block of training. Infants reached a level comparable to that of adults after twice the amount of training. As these tasks inevitably pose different cognitive and sensory demands to the two groups, in Experiments 2 and 3 we explored how varying parameters of the learning context affect dynamics of category formation. Decreasing memory demands of the task resulted in an acceleration of infants’ category formation (Experiment 2), whereas posing memory load in an implicit category learning task decelerated adults’ dynamics of category formation (Experiment 3).