Functions and Mechanisms of Intrinsic Motivations
Mammals, and humans in particular, are endowed with an exceptional capacity for cumulative learning. This capacity crucially depends on the presence of intrinsic motivations, that is, motivations that are directly related not to an organism’s survival and
- PDF / 315,638 Bytes
- 24 Pages / 439.36 x 666.15 pts Page_size
- 71 Downloads / 244 Views
Abstract Mammals, and humans in particular, are endowed with an exceptional capacity for cumulative learning. This capacity crucially depends on the presence of intrinsic motivations, that is, motivations that are directly related not to an organism’s survival and reproduction but rather to its ability to learn. Recently, there have been a number of attempts to model and reproduce intrinsic motivations in artificial systems. Different kinds of intrinsic motivations have been proposed both in psychology and in machine learning and robotics: some are based on the knowledge of the learning system, while others are based on its competence. In this contribution, we discuss the distinction between knowledge-based and competence-based intrinsic motivations with respect to both the functional roles that motivations play in learning and the mechanisms by which those functions are implemented. In particular, after arguing that the principal function of intrinsic motivations consists in allowing the development of a repertoire of skills (rather than of knowledge), we suggest that at least two different sub-functions can be identified: (a) discovering which skills might be acquired and (b) deciding which skill to train when. We propose that in biological organisms, knowledge-based intrinsic motivation mechanisms might implement the former function, whereas competencebased mechanisms might underlie the latter one.
M. Mirolli () G. Baldassarre Istituto di Scienze e Tecnologie della Cognizione, Consiglio Nazionale delle Ricerche, Rome, Italy e-mail: [email protected]; [email protected] G. Baldassarre and M. Mirolli (eds.), Intrinsically Motivated Learning in Natural and Artificial Systems, DOI 10.1007/978-3-642-32375-1 3, © Springer-Verlag Berlin Heidelberg 2013
49
50
M. Mirolli and G. Baldassarre
1 Introduction The capacity of autonomous cumulative learning demonstrated by complex organisms like mammals, and humans in particular, is astonishing. This capacity is likely to have its roots in intrinsic motivations, that is, motivations not directly related to extrinsic rewards such as food or sex, but rather to what the animal knows (curiosity, novelty, surprise) or can do (competence). Both animal and human psychologists have found evidence indicating that intrinsic motivations play an important role in animals’ behavior and learning (Berlyne 1960; Deci 1975; Deci and Ryan 1985; White 1959). Recently, the study of intrinsic motivations has been gaining increasing attention also in machine learning and robotics, as researchers in these fields have recognized that truly intelligent artificial systems need to develop their own abilities while autonomously interacting with their environment (Weng et al. 2001). The potentially open-ended complexification of a system’s skills might require the use of learning signals that are non-task specific and hence intrinsic. As a result, several computational models of intrinsically motivated learning have been proposed so far, but the study of artificial intrinsica
Data Loading...
