Agency (psychology)


In psychology, agents are goal-directed entities that are able to monitor their environment to select and perform efficient means-ends actions that are available in a given situation to achieve an intended goal. Agency, therefore, implies the ability to perceive and to change the environment of the agent, but crucially, it also entails intentionality to represent the goal-state in the future, equifinal variability to be able to achieve the intended goal-state with different actions in different contexts, and rationality of actions in relation to their goal to produce the most efficient action available. Cognitive scientists and psychologists thoroughly investigated agency attribution in humans and non-human animals, since social cognitive mechanisms as communication, social learning, imitation or theory of mind presupposes the ability to identify agents and differentiate them from inanimate objects. This ability has also assumed to have a major effect on inferential and predictive processes of the observers of agents, because agentive entities are expected to perform autonomous behavior based on their current and previous knowledge and intentions, while inanimate objects are supposed to react to external physical forces.
Although it is often confused, sensitivity to agency and the sense of agency are different concepts as the sensitivity to agency can be explained as a cognitive ability to identify agentive entities in the environment, while the sense of agency refers to the sense of control and sometimes to self-efficacy, which is a learnt belief of an individual about his or her own ability to succeed in specific situations.

Theoretical approaches of agency

According to Carey and Spelke the cognitive models explaining specific perceptual and representational abilities, for instance the models of agency recognition, can be separated into two different classes. The feature-based models of agency assume that the perceptual input of an observer consist of featural and behavioral cues that help to identify agents. Previous studies show that even very young human observers are sensitive to
However, neither of these cues are necessary and sufficient to identify an agent, since unfamiliar, novel entities like animated figures or robots without human features can elicit agency attribution in humans. Therefore, cognitive models belonging to the principle-based approaches were designed to describe how humans perceive agency assuming that the detection of agency is not a precondition, but a consequence of inferential processes about potentially agentive objects.
The theory of teleological stance proposes that from 12-month of age humans can apply the principle of rational action to determine, whether the observed entity is an agent or an inanimate object expecting that an agent behaves rationally in order to achieve its goal in a given situation. The theory assumes that the rationality principle makes observers able to relate the action, the represented goal-state and the current situational constraints to decide whether an object is an agent. For instance, if infants had learnt that an abstract, unfamiliar agent approaches another entity by jumping over an obstacle, when the obstacle had been removed, they expected a new, but highly rational behavior from the agent to approach the other entity via a straight pathway. In contrast, when infants were shown that the unfamiliar entity always made a detour when approaching its goal-object exhibiting non-justifiable behavior of jumping in the absence of an obstacle, they did not expect a rational of choice of behavior when the situational constraints changed.
These results and later empirical studies underpinned that agency recognition in humans can be explained by principle-based models rather than simple perceptual cues. As Gergely and Csibra concluded from 12-month of age humans “can take the teleological stance to interpret actions as means to goals, can evaluate the relative efficiency of means by applying the principle of rational action, and can generate systematic inferences to identify relevant aspects of the situation to justify the action as an efficient means even when these aspects are not directly visible to them".

Types of agents

It was proposed that the representation of agency can be based on the sensitivity to different abilities observed in agentive entities probably in humans and perhaps in non-human species as well. In humans, the species-specific social environment allows to identify agents either based on their intentional behavior, on their non-communicative, rational, goal-directed actions or by recognizing their communicative abilities. In non-human species, however, besides these types of input information unfamiliar agents can be identified simply on the basis of their perceptual abilities, which has context-dependent effect on their behavior even in the absence a visible goal-object that may be required to assess the efficiency of goal-approach.

Instrumental agency

According to Gergely instrumental agents are intentional agents that exhibit actions in order to realize their goal states in the environment. The recognition of instrumental agents is investigated by numerous experiments in human infants, and also in non-human apes. These studies reveal that when an agent exhibit an instrumental action it is expected by human infants to achieve its goal in an efficient manner, which is rational in terms of efforts in a given context. On the other hand it is also expected by infants that an agent should have a clear goal-state to be achieved.

Communicative agency

In contrast to instrumental agents, communicative agents are intentional agents whose actions are performed to bring about a specific change in the mental representations of the addressee, for instance by providing new and relevant information. The recognition of communicative agency may allow for the observer to predict that communicative information transfer can have a relevant effect on the behavior of the agent, even if the interacting agents and their communicative signals are unfamiliar. Communicative agents are assumed to be a subset of intentional agents, since all the communicative agents are intentional by definition, however, it is not necessary that all the intentional agents possess communicative capabilities.

Navigational agency

The construal of navigational agency is based on the assumption that Leslie’s theory on agency implies two different types of distal sensitivity; distal sensitivity in space and distal sensitivity in time. While goal-directed instrumental agents need both of these abilities to represent a goal-state in the future and achieve it in a rational and efficient manner, navigational agents are supposed to have only perceptual abilities, that is a distal sensitivity in space to avoid collision with objects in their environments. A study contrasting the ability of dogs and human infants to attribute agency to unfamiliar self-propelled object showed that dogs – unlike human infants – may lack the capability to recognize instrumental agents, however they can identify navigational agents.

Agency recognition in non-human animals

The ability to represent the efficiency of goal-directed actions of an instrumental agent may be a phylogenetically ancient core cognitive mechanism that can be found in non-human primates as well. Previous research provided evidence for this assumption showing that this sensitivity affects the expectations of cotton-top tamarins, rhesus macaques, and chimpanzees. Non-human apes are able to make inferences about the goal of an instrumental agent by taking the environmental constraints that can guide the agents’ actions into account. Moreover, it seems that non-human species like dogs can recognize contingent reactivity as an abstract of cue of agency, and respond to contingent agent significantly different in contrast to inanimate objects.