Talk by Prof. Scott Johnson

Event Date: 
Tue, 30/06/2015 - 12:00

CBC Seminars

 

Scott Johnson, from UCLA Baby Lab, Los Angeles (USA), invited by Juan Manuel Toro (LCC Group)

 

Title: Constraints on Statistical Learning in Infancy

 

Abstract: 

Statistical learning is the process of identifying patterns of probabilistic co-occurrence among stimulus features, essential to our ability to perceive the world as predictable and stable. Research on auditory statistical learning has revealed that infants use statistical properties of linguistic input to discover structure--including sound patterns, words, and the beginnings of grammar--that may facilitate language acquisition. Previous research on visual statistical learning revealed abilities to discriminate probabilities in visual sequences, leading to claims of a domain-general learning device that is available early in life, perhaps at birth. In this talk, I will present new research on visual statistical learning, rule learning, finite state grammar learning, and causal structure learning from infants, adults, and computational models. This research challenges claims of domain-generality and works toward the twin goals of understanding perceptual and memory constraints on learning and providing a mechanistic explanation of the computations underlying performance in statistical learning tasks.
 

When:   (non standard day!)  

Tuesday, June 30th, at 12:00

 

Where:

Campus de la Comunicacio UPF, C/ Roc Boronat, 132. MAP
Edifici 52 (Ed. Roc Boronat), room 52.S25

Universitat Pompeu Fabra.

 

How to get there:

Metro: Llacuna or Poblenou <L4>, Glories <L1>, Tram: Ca l’aranyó [T4]