Adapting exploration to time horizons across development

• In Event: Poster Session 03
  In Poster Session: PS 03 Topic - Cognitive Processes

Abstract

To maximize rewards in the long term, individuals must adapt their exploration to time horizons, exploring more when horizons are long and less when horizons are short. While adults and adolescents adapt to time horizons (Somerville et al., 2017; Wilson et al. 2014), it is unclear when and how young children begin to do so. Children undergo a major cognitive transition in the 5-6-year period, becoming increasingly proactive in their ability to maintain goals and anticipate future needs. This transition may support the emergence of adaptive exploration. Furthermore, it is unclear how individuals explore under ambiguous time horizons, which are prevalent in daily life. Though ambiguous horizons offer no clear adaptive strategy, ambiguity is aversive and increases preference for more certain options (Ikink et al. 2019) and may thus lead to low levels of exploration across ages.

In the current laboratory study, we examined adaptive exploration in response to different time horizons in younger children (mean= 5.5 yrs, range: 5-6 yrs, N= 43), older children (mean= 11.5 yrs, range: 11-12 yrs, N= 40), and adults (mean= 19.4 yrs, range: 18-31 yrs, N= 49). In a Simplified Horizons Task (adapted from Wilson et al. 2014; see Figure 1), participants selected among Known and Unknown reward options under Long, Short, and Ambiguous time horizons. Participants’ goal was to maximize their rewards. Our main outcome of interest was whether participants chose to explore the Unknown option in their first choice of each trial (First Choice Explore), a measure of exploration. Additionally, we assessed proactive control using the AX-Continuous Performance Task (Gonthier et al. 2019) to examine individual differences in adaptive exploration.

Individuals overall adapted to time horizons, exploring more in Long compared to in Short Horizons, Est∆=.915 [.870, .951], z= 8.90, p<.001. However, adaptation differed significantly across age (Figure 1), ∆X2 (2)=30.45, p<.001: Older Children and Adults adapted to time horizons, Older Children: Est∆= .962 [.926, .983], z= 8.43, p<.001 and Adults: Est∆= .936 [.883, .971], z=7.25, p<.001, while Younger Children only did so marginally, Est∆= .651 [.487,.786], z= 1.84, p=.066. Adaptation varied significantly across individuals within age groups, ∆X2 (2)= 20.9, p< .001, reflective of individual differences that may be explained by proactive control. (Psychometric considerations in the measure of proactive control are being evaluated and individual differences results will be forthcoming). Under Ambiguous Horizons, exploration differed across age groups, F(2, 124)= 6.65, p= .002: Older Children and Adults explored significantly below chance (.25), Older Children: Est= .058, SE=.075, t= -2.65, p =.001, and Adults: Est= .005, SE= .069, t= -3.53, p < .001, while Younger Children explored significantly above chance, Est=.509, SE=.05, t=5.018, p<.001. (Figure 2).

The current findings demonstrate that adaptation to time horizons emerges around 5 - 6 years and, as children develop, they decrease their tendencies to explore under short and ambiguous time horizons. This reduction in exploration highlights a developmental shift that may lead to less learning but more adaptive decision-making.

Authors

• Winnie Zhuang, University of California - Davis
  Presenting Author

• Jesse Niebaum, University of California - Davis
  Non-Presenting Author

• Yuko Munakata, University of California - Davis
  Non-Presenting Author