Intentional action: from anticipation to goal-directed behavior

Abstract

Humans and other animals are able to guide their actions toward the realization of their own goals, both proximal and distal. Recently, cognitive neuroscientists, biologists and psychologists have begun unraveling, from diVerent but converging perspectives, the organization of goaldirected, intentional action in terms of (brain, computational) structures and mechanisms. Converging evidence indicates that several cognitive capabilities across the individual and social domains, including action planning and execution, understanding others’ intentions, cooperation and imitation are essentially goal-directed. For example, goal representations have a crucial role in the planning and control of action, and action understanding and imitation are performed at the goal rather than at the movement level (Iacoboni et al., 2005; Wohlschlager, Gattis,& Bekkering, 2003). Moreover, it has been shown that these apparently unrelated abilities and others which were believed to be in the realm of abstract thought, such as language understanding, share common representational structures and mechanisms in the brain and involve signiWcant use of the motor system (Fadiga, Craighero, Buccino, & Rizzolatti, 2002; Pulvermuller, 1999). Recent studies have revealed the crucial role of canonical and mirror neurons (Rizzolatti & Craighero, 2004; Rizzolatti et al., 1988) and internal forward models (Kawato, 1999; Wolpert & Ghahramani, 2004) in most of the aforementioned tasks. These studies indicate a close link between (socio)cognitive abilities and situated action, and are shedding light on their neural underpinnings and mechanisms. Combined theoretical, empirical and computational research is revealing that during all the aforementioned activities, the motor system is highly engaged in anticipatory, simulative and generative processes. From a speculative point of view, a case could be made that the same predictive mechanisms provide both a “linkage with the future” required for taking goal-directed action, and a “linkage with others” required to act socially (Pezzulo, 2008; Pezzulo and Castelfranchi, 2007). Another particularly intriguing aspect of these Wndings is the revelation that the two domains of perception and action, which are traditionally kept separate in the analysis of cognitive systems, cannot be disentangled. Again, anticipatory phenomena, such as expectations used to control action, and simulative processes, play a major role in the perception–action linkage, and give rise to ideomotor neural codes that relate actions and their eVects (Hommel, Musseler, Aschersleben, & Prinz, 2001). This seems to happen at multiple levels and encompasses sensorimotor and higher-level cognitive tasks. Overall, these recent studies (and others) have led to a profound rethinking of basic concepts in cognitive and behavioral sciences, and a common theoretical view (a motor-based (or action-based) view of cognition) is emerging across disciplines (Jeannerod, 2006). In this context, the abilities of action execution, its planning and understanding of others’ intentions are all described as essentially goal-directed and served by the same representations, which are action-oriented and involve deeply the motor apparatus. Simulative theories of cognition add to G. Pezzulo (&) Istituto di Linguistica Computazionale “Antonio Zampolli”, CNR, Via Giuseppe Moruzzi, 1, 56124 Pisa, Italy e-mail: giovanni.pezzulo@cnr.it