Poster #10 - Infants’ brain response to predictable stimuli is weaker but more synchronized

• In Event: 1-061 - Poster Session 02
  In Poster Session: PS 02 Section - Attention, Learning, Memory

Integrative Statement

Prediction about upcoming events can be considered a defining feature of development (Piaget, 1952). While predictions allow infants to orient their attention towards surprising events (Hunter & Ames, 1988; Kidd et al., 2014) and facilitates learning (Stahl and Feigenson, 2015), little is known about how predictions shape the developing brain. In adults, it is believed that predictions employ top-down modulation to compare sensory input against the prediction (Rao & Ballard, 1999; Friston, 2005). Similar to what is found in adults, recent work has suggested that predictions are evident in sensory cortices in infancy (Emberson et al., 2015; Kouider et al., 2015). However, these previous studies only investigated one type of prediction (i.e., arising from learning an audiovisual association and looking at visual prediction based on an auditory cue) and didn’t provide direct evidence that disparate regions of the brain were involved in the generation of prediction.
Here, we investigate prediction in a new context, which requires integration of stimuli across longer time scales, and measure the impact of prediction on neural responses in the sensory and associative (higher-level) cortices in the infant brain using whole-head fNIRS (74 channels). Consistent with the framework of predictive coding, we hypothesized that neural response in sensory cortices would be attenuated for predictable stimuli, suggesting that prediction improves the efficiency of perception of these stimuli through the use of top-down information from associative cortices. Additionally, we hypothesized that this cortical attenuation to predictable stimuli will be accompanied by increased functional connectivity between frontal and posterior regions (i.e., synchronized neural activity).
We recruited 30 typically-developing full-term infants at 6 months. We measured infants’ cortical correlates of predictability by contrasting their neural response to Predictable versus Unpredictable sequences of audio-visual events that they learn within the experimental session (Fig. 1). In the Predictable condition, the temporal order of the stimuli was constant across trials, while in the Unpredictable condition, the temporal order varied from trial to trial. Thus, in the Predictable condition, once the temporal pattern is learned, the infant would be able to predict which stimuli should appear next. Such prediction will be impossible to make in the Unpredictable condition.
As hypothesized, we found that in the Unpredictable condition infants had a stronger neural response in both frontal and posterior (occipital and temporal) regions than in the Predictable condition (Fig. 2). Predictions modulate neural response to stimuli, across disparate, sensory specific and cross-modal, associative brain regions and thus employs long-range connectivity. Indeed, we found greater functional connectivity in the Predictable condition.
Ongoing work is investigating these effects in preterm infants (born <34 weeks). Previous work has found weaker evidence for prediction in preterm infants (Emberson et al., 2017). We thus expect preterm infants to exhibit smaller differences between the Unpredictable and Predictable conditions.

Authors

• Sagi Jaffe-Dax, Princeton University
  Presenting Author

• Vikranth Rao Bejjanki, Hamilton College
  Non-Presenting Author

• Lauren Emberson, Princeton University
  Non-Presenting Author