Completion of the Infeasible Actions of Others: Goal Inference by Dynamical Invariant

To help another person, we need to infer his or her goal and intention and then perform the action that he or she was unable to perform to meet the intended goal. In this study, we investigate a computational mechanism for inferring someone's intention and goal from that person's incomplete action to enable the action to be completed on his or her behalf. As a minimal and idealized motor control task of this type, we analyzed single-link pendulum control tasks by manipulating the underlying goals. By analyzing behaviors generated by multiple types of these tasks, we found that a type of fractal dimension of movements is characteristic of the difference in the underlying motor controllers, which reflect the difference in the underlying goals. To test whether an incomplete action can be completed using this property of the action trajectory, we demonstrated that the simulated pendulum controller can perform an action in the direction of the underlying goal by using the fractal dimension as a criterion for similarity in movements.

MapGo: Model-Assisted Policy Optimization for Goal-Oriented Tasks

In Goal-oriented Reinforcement learning, relabeling the raw goals in past experience to provide agents with hindsight ability is a major solution to the reward sparsity problem. In this paper, to enhance the diversity of relabeled goals, we develop FGI (Foresight Goal Inference), a new relabeling strategy that relabels the goals by looking into the future with a learned dynamics model. Besides, to improve sample efficiency, we propose to use the dynamics model to generate simulated trajectories for policy training. By integrating these two improvements, we introduce the MapGo framework (Model-Assisted Policy optimization for Goal-oriented tasks). In our experiments, we first show the effectiveness of the FGI strategy compared with the hindsight one, and then show that the MapGo framework achieves higher sample efficiency when compared to model-free baselines on a set of complicated tasks.

I Know You Know I’m Signaling: Novel gestures are designed to guide observers’ inferences about communicative goals

For a gesture to be successful, observers must recognize its communicative purpose. Are communicators sensitive to this problem and do they try to ease their observer’s inferential burden? We propose that people shape their gestures to help observers easily infer that their movements are meant to communicate. Using computational models of recursive goal inference, we show that this hypothesis predicts that gestures ought to reveal that the movement is inconsistent with the space of non-communicative goals in the environment. In two gesture-design experiments, we find that people spontaneously shape communicative movements in response to the distribution of potential instrumental goals, ensuring that the movement can be easily differentiated from instrumental action. Our results show that people are sensitive to the inferential demands that observers face. As a result, people actively work to help ensure that the goal of their communicative movement is understood.

Constituents’ Inferences of Government Goals Explain the Relationship Between Political Party and Belief in COVID-19 Misinformation: A Cross-Sectional Survey of Twitter Users Who Follow their State’s Public Health Twitter Account (Preprint)

BACKGROUND Amidst the widespread global COVID-19 pandemic, social media have played a pivotal role in the circulation of health information. Public health agencies often use Twitter as a tool to disseminate and amplify the propagation of such information [1], but exposure to local government-endorsed COVID-19 public health information does not make one immune to believing in misinformation. Moreover, not all health information on Twitter is accurate, and some users may believe misinformation and disinformation just as much as those who endorse more accurate information [2]. This situation is complicated when considering the unfortunate reality that elected officials may be promoting misinformation in pursuit of their other political agendas, like downplaying the need for COVID-19 restrictions to promote their reelection bid [3]. The politicized and polarized nature of information surrounding COVID-19 on social media in the U.S. has fueled a concomitant COVID-19 social media infodemic [4-6]. As such, because pre-existing political beliefs can both facilitate and hinder persuasion [7,8], goal understanding processes are likely at work in the belief of COVID-19 misinformation for Twitter users, such that the valence of users’ goal inferences for their local government agencies likely impact the extent to which they believe state government-endorsed COVID-19 information disseminated via social media. OBJECTIVE The present investigation sheds light on the cognitive processes of goal understanding that underlie the relationship between partisanship and belief in health misinformation. We investigate how Twitter users’ goal inference valence of local government’s COVID-19 efforts predicts their beliefs in COVID-19 misinformation as a function of their political party. METHODS We conducted an online cross-sectional survey of U.S. Twitter users who followed their state’s official department of public health Twitter (n=258) between August 10 and December 23, 2020. Local government goal inferences, demographics, and COVID-19 misinformation were measured. State political affiliation was controlled. RESULTS Participants from all 50 states were in the sample. Results revealed an interaction between political party affiliation and goal inference valence on belief in covid misinformation, R2∆ = .04, F(8,249) = 4.78, p < .001, such that positive goal inference valence predicted increased belief in COVID-19 misinformation for Republicans, β=.47, t(249) = 2.59, p = 0.01 but not Democrats, β= .07, t(249) = 0.84, p = 0.40. CONCLUSIONS Results reveal that positive inferences about local government’s COVID-19 efforts can accelerate beliefs in misinformation for Republican-identifying constituents. Republicans’ inferences that their local government has positive intentions may make republican constituents more vulnerable to republican-endorsed COVID-19 misinformation. In other words, accurate COVID-19 transmission knowledge has been driven by constituents' sentiment about politicians rather than science. This work stresses the need for health campaigns to be sensitive to the preexisting political affiliation of their target audience when constructing persuasive health messages.

Boolean Games: Inferring Agents' Goals Using Taxation Queries

In Boolean games, each agent controls a set of Boolean variables and has a goal represented by a propositional formula. We study inference problems in Boolean games assuming the presence of a PRINCIPAL who has the ability to control the agents and impose taxation schemes. Previous work used taxation schemes to guide a game towards certain equilibria. We present algorithms that show how taxation schemes can also be used to infer agents' goals. We present experimental results to demonstrate the efficacy our algorithms. We also consider goal inference when only limited information is available in response to a query.

Expanding Advice Response Theory by Incorporating Goal Inference: College Students’ Perceptions of Parental Exercise Advice

We expanded Advice Response Theory (ART) by proposing that recipient perceptions of advisor characteristics can be distal (e.g., parenting style) and proximal (e.g., goal inference). We examined how the recipient’s inference of the advisor’s goals (confirmation, change, and novelty) mediates associations between distal characteristics and message feature evaluations and outcomes. As predicted, positive associations occurred between perceptions of parenting style, confirmation goal inference, advice efficacy and positive facework, and desirable advice outcomes. Counter to predictions, inferring change and novelty goals did not have uniformly undesirable effects. An inference of the change goal was associated with higher efficacy ratings and the recipient changing plans following the conversation. Our findings support conceptualizing ART advisor characteristics as distal and proximal, and advisor goal inference as a relevant proximal characteristic.

Bridging the Gap between Observation and Decision Making: Goal Recognition and Flexible Resource Allocation in Dynamic Network Interdiction

Goal recognition, which is the task of inferring an agent’s goals given some or all of the agent’s observed actions, is one of the important approaches in bridging the gap between the observation and decision making within an observe-orient-decide-act cycle. Unfortunately, few researches focus on how to improve the utilization of knowledge produced by a goal recognition system. In this work, we propose a Markov Decision Process-based goal recognition approach tailored to a dynamic shortest-path local network interdiction (DSPLNI) problem. We first introduce a novel DSPLNI model and its solvable dual form so as to incorporate real-time knowledge acquired from goal recognition system. Then a Markov Decision Process-based goal recognition model along with its dynamic Bayesian network representation and the applied goal inference method is proposed to identify the evader’s real goal within the DSPLNI context. Based on that, we further propose an efficient scalable technique in maintaining action utility map used in fast goal inference, and develop a flexible resource assignment mechanism in DSPLNI using knowledge from goal recognition system. Experimental results show the effectiveness and accuracy of our methods both in goal recognition and dynamic network interdiction.