Consider a wildlife photographer that has just entered a rainforest that she has never visited. Looking for a good spot for animal photos, she can spend all her time in the first hideout that she found, slowly learning which animals visit that spot. Alternatively, she can consider other locations, which are potentially better but might also be worse. To identify these better locations she needs to leave her hideout and walk further into the forest, thus missing the opportunity to learn more about the qualities of her first hideout. How should she explore the forest? How does she explore it? Here we describe the computational principles and algorithms underlying exploration in the field of Machine Learning and discuss their relevance to human behavior.
Paper of the month
Loewenstein’s Lab: Exploration: from machines to humans
Current Opinion in Behavioral Sciences, Volume 35, 2020, Pages 104-111, ISSN 2352-1546, (2020)