Towards computational models of intention detection and intention prediction

Intention recognition is one of the core components of mindreading, an important process in social cognition. Human beings, from age of 18 months, have been shown to be able to extrapolate intentions from observed actions, even when the performer failed at achieving the goal. Existing accounts of intention recognition emphasize the use of an intent (plan) library, which is matched against observed actions for recognition. These therefore cannot account for recognition of failed sequences of actions, nor novel actions. In this paper, we begin to tackle these open questions by examining computational models for components of human intention recognition, which emphasize the ability of humans to detect and identify intentions in a sequence of observed actions, based solely on the rationality of movement (its efficiency). We provide a high-level overview of intention recognition as a whole, and then elaborate on two components of the model, which we believe to be at its core, namely, those of intention detection and intention prediction. By intention detection we mean the ability to discern whether a sequence of actions has any underlying intention at all, or whether it was performed in an arbitrary manner with no goal in mind. By intention prediction we mean the ability to extend an incomplete sequence of actions to its most likely intended goal. We evaluate the model, and these two components, in context of existing literature, and in a number of experiments with more than 140 human subjects. For intention detection, our model was able to attribute high levels of intention to those traces perceived by humans as intentional, and vice versa. For intention prediction as well, our model performed in a way that closely matched that of humans. The work highlights the intimate relationship between the ability to generate plans, and the ability to recognize intentions.     

Construction of a conditional probability learning model and information theoretical evaluation of its efficiency

A learning model is introduced based on the simple idea that a learning subject has two hypotheses about the results (the outputs of the environment) caused by his actions and calculates the a posteriori probabilities of the hypotheses. Introducing a randomized decision rule of actions related to the probabilities, it is shown that the model is equivalent to Luce's beta model, while these models are derived from completely different considerations. Introducing an information theoretic measure and martingale theory, the asymptotic behavior of the model is examined. Using the information theoretic measure, the effect of memory function on the learning behavior is evaluated. Finally, the model is extended in such a way that the learning subject has more than two hypotheses or a continuum of hypotheses. The learning behaviors of the extended models are discussed.     

Pack behavior

We analyze a sequential decision model, where every decision maker has her own preferences over actions. Some actions have positive Externalities if a large enough population has previously chosen them, and those Externalities may lead a decision-maker to ignore her own preferred actions to benefit from those Externalities. We give a fairly general sufficient condition such that, on almost every equilibrium path, every decision maker eventually prefers to benefit from group Externalities. We show that this situation, called pack behavior, is observationally equivalent to a herd behavior, but it is triggered by positive Externalities generated by group decision, instead of standard belief-based decisions. We also show that this notion is not ex-ante Pareto-efficient in general.     

Konzeption zum Umgang mit rückfallgef?hrdeten Sexualstraft?tern (KURS NRW)

The concept of the management of sexual offenders at risk for recidivism in Northrhine-Wesfalia (Konzeption zumUmgang mitrückfallgef?hrdetenSexualstraft?tern in Nordrhein-Westfalen, KURS NRW) came into effect on February 1, 2010 with a joint circular of the Ministries of Justice, Health and the Interior. Thereby Northrhine-Westfalia has created a systematic for monitoring sexual offenders at risk for recidivism, comparable to those already or in the meantime developed in other federal states in Germany. This article initially describes the procedural nature underlying the KURS concept, regarding mutual information routing and collaboration between the parties involved, i.e. justice, hospital order treatment, police and if necessary others as well. In addition, the framework in which KURS is ingrained, will be shown. Based on case reports this article will finally explain how actions and reactions will be specified in case conferences. Within the systematic KURS it is important that all participants reach all relevant information expeditiously, thus being able to react within their own jurisdiction. When planning actions it is essential to carefully consider the individual case parameters. Within the case conferences it remains a strength of the KURS concept that different perspectives of the case and the offender are bundled together. Ideally this and the case-specific actions will help to avoid similar offences to those committed previously.     

“Should I or shouldn’t I?” Imitation of undesired versus allowed actions from peer and adult models by 18- and 24-month-old toddlers

Imitation is a common way of acquiring novel behaviors in toddlers. However, little is known about toddlers’ imitation of undesired actions. Here we investigated 18- and 24-month-olds’ (N = 110) imitation of undesired and allowed actions from televised peer and adult models. Permissiveness of the demonstrated actions was indicated by the experimenter’s response to their execution (angry or neutral). Analyses revealed that toddlers’ imitation scores were higher after demonstrations of allowed versus undesired actions, regardless of the age of the model. In agreement with prior research, these results suggest that third-party reactions to a model’s actions can be a powerful cue for toddlers to engage in or refrain from imitation. In the context of the present study, third-party reactions were more influential on imitation than the model’s age. Considering the relative influence of different social cues for imitation can help to gain a fuller understanding of early observational learning.     

The response model of moral disgust

The philosophical debate over disgust and its role in moral discourse has focused on disgust’s epistemic status: can disgust justify judgments of moral wrongness? Or is it misplaced in the moral domain—irrelevant at best, positively distorting at worst? Correspondingly, empirical research into disgust has focused on its role as a cause or amplifier of moral judgment, seeking to establish how and when disgust either causes us to morally condemn actions, or strengthens our pre-existing tendencies to condemn certain actions. Both of these approaches to disgust are based on a set of assumptions that I call, in what follows, the evidential model of disgust. This paper proposes an alternative model, which I call the response model. Instead of looking at disgust as a cause and justification of judgments of moral wrongness, I will argue that disgust is better understood as a response to wrongness. More precisely, I argue that disgust is a response to norm violations, and that it is (sometimes) a fitting response insofar as norm violations are potentially contagious and therefore pose a threat to the stability and maintenance of norms.     

Inferring the intentional states of autonomous virtual agents

Inferring the mental states of other agents, including their goals and intentions, is a central problem in cognition. A critical aspect of this problem is that one cannot observe mental states directly, but must infer them from observable actions. To study the computational mechanisms underlying this inference, we created a two-dimensional virtual environment populated by autonomous agents with independent cognitive architectures. These agents navigate the environment, collecting “food” and interacting with one another. The agents’ behavior is modulated by a small number of distinct goal states: attacking, exploring, fleeing, and gathering food. We studied subjects’ ability to detect and classify the agents’ continually changing goal states on the basis of their motions and interactions. Although the programmed ground truth goal state is not directly observable, subjects’ responses showed both high validity (correlation with this ground truth) and high reliability (correlation with one another). We present a Bayesian model of the inference of goal states, and find that it accounts for subjects’ responses better than alternative models. Although the model is fit to the actual programmed states of the agents, and not to subjects’ responses, its output actually conforms better to subjects’ responses than to the ground truth goal state of the agents.     

Change processes in development: The concept of coactive scaffolding

The concept of scaffolding is generally invoked to refer to the ways in which a more expert individual assists a child by performing a part a task or by otherwise directing or supporting a child's task-related actions. A coactive systems model of development provides a framework for examining other ways in which person-environment relations may scaffold development. From a coactive systems view, the unit of analysis for understanding development is the coactive person-environment system. Within such a system, although individual actors exert control over their actions, thoughts and feelings, action is the product of coactions among each element of the system over time. From this view, coactive scaffolding refers to any process outside of an individual's direct control that functions to direct individual action toward novel or higher-order forms. Three broad categories (and subtypes) of coactive scaffolding are proposed and illustrated: ecological scaffolding, social scaffolding, and self-scaffolding.     

Action,prediction, and temporal awareness

The brain needs to track changes in the relation between action and effect. In two experiments, participants made voluntary keypress actions. In an adaptation phase, these actions were followed after a fixed interval by a tone. During a subsequent test phase, the duration of the interval was unexpectedly changed. We used time perception as an implicit marker of the experience of participants' control over the effect, and confirmed a temporal binding between actions and effects. On test trials, participants perceived tones to occur as shifted towards their time of occurrence in the preceding adaptation phase. Therefore, the perceived time of a tone was partly based on learning of an internal prediction, rather than on the time of actual sensory input. This predictive model is rapidly updated over a few trials (Experiment 1), and requires attention to the tones (Experiment 2). The brain learns action–effect relations. This predictive learning influences the perception of effects, and underlies some temporal illusions associated with action.     

Deferred imitation in 18-month-olds from two cultural contexts: The case of Cameroonian Nso farmer and German-middle class infants

Imitative learning has been described in naturalistic studies for different cultures, but lab-based research studying imitative learning across different cultural contexts is almost missing. Therefore, imitative learning was assessed with 18-month-old German middle-class and Cameroonian Nso farmer infants – representing two highly different eco-cultural contexts associated with different cultural models, the psychological autonomy and the hierarchical relatedness – by using the deferred imitation paradigm. Study 1 revealed that the infants from both cultural contexts performed a higher number of target actions in the deferred imitation than in the baseline phase. Moreover, it was found that German middle-class infants showed a higher mean imitation rate as they performed more target actions in the deferred imitation phase compared with Cameroonian Nso farmer infants. It was speculated that the opportunity to manipulate the test objects directly after the demonstration of the target actions could enhance the mean deferred imitation rate of the Cameroonian Nso farmer infants which was confirmed in Study 2. Possible explanations for the differences in the amount of imitated target actions of German middle-class and Cameroonian Nso farmer infants are discussed considering the object-related, dyadic setting of the imitation paradigm with respect to the different learning contexts underlying the different cultural models of learning.     

Eudaimonist Virtue Ethics and Right Action: A Reassessment

My question in this paper concerns what eudaimonist virtue ethics (EVE) might have to say about what makes right actions right. This is obviously an important question if we want to know what (if anything) distinguishes EVE from various forms of consequentialism and deontology in ethical theorizing. The answer most commonly given is that according to EVE, an action is right if and only if it is what a virtuous person would do in the circumstances. However, understood as a claim about what makes particular actions right, this is not especially plausible. What makes a virtuous person??s actions right must reasonably be a matter of the feature, or features, which she, via her practical wisdom, appreciates as ethically relevant in the circumstances, and not the fact that someone such as herself would perform those actions. I argue that EVE instead should be understood as a more radical alternative in ethical philosophy, an alternative that relies on the background assumption that no general account or criterion for what makes right actions right is available to us: right action is simply too complex to be captured in a ??finite and manageable set of??moral principles?? (McKeever and Ridge, Principled ethics, Oxford University Press, 2006, p. 139). This does not rule out the possibility that there might be some generalizations about how we should act which hold true without exception. Perhaps there are some things which we must never do, as well as some features of the world which always carry normative weight (even though their exact weight may vary from one context to another). Still, these things are arguably few and far between, and what we must do to ensure that we reliably recognize what is right in particular situations is to acquire practical wisdom. Nothing short of that could do the job.     

Promotionalism, Motivationalism and Reasons to Perform Physically Impossible Actions

In this paper I grant the Humean premise that some reasons for action are grounded in the desires of the agents whose reasons they are. I then consider the question of the relation between the reasons and the desires that ground them. According to promotionalism, a desire that p grounds a reason to φ insofar as A’s φing helps promote p. According to motivationalism a desire that p grounds a reason to φ insofar as it explains why, in certain circumstances, A would be motivated to φ. I then give an argument favouring motivationalism, namely that promotionalism entails that agents have reasons to perform physically impossible actions, whereas motivationalism entails that there are no such reasons. Although this is a version of the ‘Too Many Reasons’ objection to promotionalism, I show that existing responses to that problem do not transfer to the case of reasons to perform physically impossible actions. In the penultimate section I consider and reject some objections to motivationalism made by promotionalists. The conclusion is that Humeans about reasons for action should prefer motivationalism.     

The idea of the absurd and the moral decision. Possibilities and limits of a physician's actions in the view of the absurd

In reference to two central concepts of Albert Camus' philosophy, that is, the absurd and the rebellion, this article examines to what extent hisThe Plague is of interest to medical ethics. The interpretation of this novel put forward in this article focuses on the main character of the novel, the physician Dr. Rieux. For Rieux, the plague epidemic, as it is described in the novel, implies an unquestioning commitment to his patients and fellow men. According to Camus this epidemic has to be understood as a symbol of the absurd. Unable to base his actions on a Christian, metaphysical value system, Rieux sees his commitment as a continuous rebellion against the fact of the absurd, which opposes him in the form of evil, suffering and death. As a physician, Rieux is therefore forced to adjust his actions to life in its immediacy, that is, the suffering of his patients. In this article, it will be shown that Rieux's attention to the “immediate” is of particular interest to medical ethics: Theother person in need, rather thanmy moral convictions, sets the norm.     

Modulation of motor area activity during observation of unnatural body movements

The mirror neuron system (MNS) is activated when observing the actions of others. However, it remains unclear whether the MNS responds more strongly to natural bodily actions in the observer’s motor repertoire than to unnatural actions. We investigated whether MNS activity is modulated by the unnaturalness of an observed action by inserting short pauses in the middle of the action (0, 2, and 6 pauses; no-pause, pause-1, and pause-2 conditions, respectively). The results indicated that the number of pauses significantly affected motor area activity. Subsequent analyses revealed significant differences between the pause-1 and pause-2 conditions (P < 0.01), as well as the no-pause and pause-2 conditions (P < 0.05). There was significant activation in the pause-1 condition (P < 0.001), while significant deactivation was observed in the pause-2 condition (P < 0.05). These results indicate that MNS activity is modulated by the kinematic characteristics of the observed action. We suggest the possibility that a slightly deviated action may enhance the MNS activity during action observation, while a highly unnatural action would lead to a strong attenuation (deactivation) of activity in the MNS.     

Conceptually congruent actions can promote thought

Can action support thought? Previous work suggests that it can. Here, we examined whether actions that are conceptually congruent with thinking facilitate thinking and whether direct action facilitates performance. We found that young children performed addition, a discrete one-to-one math task, better when using discrete one-to-one actions that matched the number of objects than when using discrete actions that matched the number of sums to be added. They performed number line estimation, a continuous math task, better when using a continuous action in which the time and distance of the action were commensurate with the quantity to be estimated, than when using a discrete action that marked a proportional distance. Action congruence facilitated performance beyond spatial congruence. Furthermore, direct manipulation led to better performance than mediated manipulation. Finding advantages of congruent mappings of thought to action supports the Spraction Theory, which asserts that thought is internalized action, and that re-externalizing thought through congruent actions facilitates thought.     

Subliminal action priming modulates the perceived intensity of sensory action consequences

The sense of control over the consequences of one’s actions depends on predictions about these consequences. According to an influential computational model, consistency between predicted and observed action consequences attenuates perceived stimulus intensity, which might provide a marker of agentic control. An important assumption of this model is that these predictions are generated within the motor system. However, previous studies of sensory attenuation have typically confounded motor-specific perceptual modulation with perceptual effects of stimulus predictability that are not specific to motor action. As a result, these studies cannot unambiguously attribute sensory attenuation to a motor locus. We present a psychophysical experiment on auditory attenuation that avoids this pitfall. Subliminal masked priming of motor actions with compatible prime–target pairs has previously been shown to modulate both reaction times and the explicit feeling of control over action consequences. Here, we demonstrate reduced perceived loudness of tones caused by compatibly primed actions. Importantly, this modulation results from a manipulation of motor processing and is not confounded by stimulus predictability. We discuss our results with respect to theoretical models of the mechanisms underlying sensory attenuation and subliminal motor priming.     

Living,resisting, and playing the part of athlete: Narrative tensions in elite sport

ObjectivesTo explore: (i) How elite and professional sport culture might steer individuals towards particular stories, identities, and actions; (ii) How athletes navigate or respond to these cultural pressures.DesignCross-sectional qualitative methodology.MethodNarrative interviews and focus groups with 21 elite and professional athletes followed by a narrative analysis of structure and form.ResultsAthletes demonstrated one of three processes. Individuals who live the part of athlete story their life and act in ways that conform to a culturally dominant performance narrative. Here, identity is foreclosed, relationships sacrificed in the pursuit of success, and long-term wellbeing threatened. Over time, alternative narrative types may provoke moral reflection on their story and actions. Individuals who resist the part of athlete sustain a life story and identity that deviates from the performance narrative, drawing on alternative narrative types. Their resistance is typically overt as they publicly demonstrate actions that align with their multidimensional story. Individuals who play the part of athlete modify their story and actions depending on sociocultural context. These individuals covertly maintain a multidimensional life story, but silence this story when powerful others require performance stories.ConclusionsAlthough some elite/professional athletes' life stories revolve around performance outcomes, this is not a prerequisite for excellence. Other athletes achieve excellence while sustaining a multidimensional life story and identity. To do so, they navigate a culture that expects a performance focus, through overt resistance or covertly manipulating their public stories and actions.     

Revisiting the Cultural Dope

This essay focuses on the ??cultural dope,?? an ironic reference in Harold Garfinkel??s Studies in Ethnomethodology to the rule-following actor in conventional sociological theories. In the nearly half-century since the publication of that book, the ??cultural dope?? has been incorporated into numerous criticisms of ??models of man?? in the human sciences. Garfinkel??s account appeals to many writers because it seems to present an alternative picture of the actor: an individual who is self-aware, reflective, and skilled in the conduct of daily affairs. A problem with such a generalized picture of the actor is that it may seem to encourage uncritical acceptance of whatever ??the public?? (or a broad segment of the public) happens to believe or support. This paper revisits Garfinkel??s account of the cultural dope, and contrasts ??conservative?? and ??radical?? readings of what Garfinkel does with that figure. The ??conservative?? reading leaves the edifice of a social-structural model largely intact, and provides an alternative, more complex, picture of individual action than that of a cultural dope. The ??radical?? reading places relevant social structures in a dependent relation to the contingencies of action, and thus destabilizes the very theoretical edifice that sets up the problem of how to integrate individual actions with stable social structures. In line with the ??radical?? reading, this paper suggests that Garfinkel creates serious difficulty for any generalized ??model of man,?? regardless of whether it portrays the individual as active or passive, well-informed or ignorant, or reflexive or not.     

Statements of Method and Teaching: The Case of Socrates

In this paper, I ponder the question of whether Socrates follows a method of investigation — the method of hypothesis — which he advocates in Plato's Phaedo. The evidence in the dialogue suggests that he does not follow the method, which raises additional questions: If he fails to do so, why does he articulate the method? Does his statement of method affect his actions or is it mainly forgotten? Although Socrates is a fictional character, his actions in the Phaedo suggests questions about the function of espoused methods in actual situations.