Cookies on this website
We use cookies to ensure that we give you the best experience on our website. If you click 'Continue' we'll assume that you are happy to receive all cookies and you won't see this message again. Click 'Find out more' for information on how to change your cookie settings.

© 2016 IEEE. Humans, other animals, and artificial systems solve problems posed by their physical environment. Biological agents start with a flexible cognitive architecture but limited abilities, and acquire competences by different combinations of maturation, growth and experience (both social and individual), and some designers of artificial systems try to emulate such processes. By examining the processes leading to the construction of competence in biological agents we hope to inspire and support ideas relevant to the design of intelligent agents in the future. Here we report a study of the emergence of physical problem-solving in human infants aged 15 to 24 months and in both tool-using and non-tool-using corvids. One goal is to examine whether tool-using competence was indicative of general physical intelligence, in which case tool using agents should show greater competence both in tool-related and non-tool related tasks. Another is to test whether in these disparate organisms, experience with combining objects in contexts other than functional tool use (e.g. play) can causally facilitate the emergence of tool use competence. We used a battery of tasks including tool vs. non-tool dependent extractions of target objects. All subjects were initially exposed to a mixture of objects and allowed to interact freely with them. Preliminary results show that while there are strong inter-species differences in tool-related competence, we found no major differences in tasks where tools were not involved. There were strong individual differences within each species, but age effects were only found in human infants, reflecting differences in life-history: Although crows can live for several decades, they reach adulthood (including asymptotic competence in many indices) much faster than humans. From a robotics perspective, the differences reveal available choices regarding the tradeoff of learning vs. performing, or exploration vs. exploitation of skills. The relation between object combinations and individual performance at the tool tasks is still under analysis.

Original publication

DOI

10.1109/DEVLRN.2016.7846797

Type

Conference paper

Publication Date

07/02/2017

Pages

101 - 102