Putting perception into action: Inverse optimal control for continuous psychophysics

Psychophysical methods are a cornerstone of psychology, cognitive science, and neuroscience where they have been used to quantify behavior and its neural correlates for a vast range of mental phenomena. Their power derives from the combination of controlled experiments and rigorous analysis through signal detection theory. Unfortunately, they require many tedious trials and preferably highly trained participants. A recently developed approach, continuous psychophysics, promises to transform the field by abandoning the rigid trial structure involving binary responses and replacing it with continuous behavioral adjustments to dynamic stimuli. However, what has precluded wide adoption of this approach is that current analysis methods recover perceptual thresholds, which are one order of magnitude larger compared to equivalent traditional psychophysical experiments. Here we introduce a computational analysis framework for continuous psychophysics based on Bayesian inverse optimal control. We show via simulations and on previously published data that this not only recovers the perceptual thresholds but additionally estimates subjects' action variability, internal behavioral costs, and subjective beliefs about the experimental stimulus dynamics. Taken together, we provide further evidence for the importance of including acting uncertainties, subjective beliefs, and, crucially, the intrinsic costs of behavior, even in experiments seemingly only investigating perception.

Not Quite All in Our Head: Intervention is a Better Predictor of tDCS Mind-Wandering Effects than Subjective Beliefs About Experimental Results

Background: Establishing adequate blinding for non-invasive brain stimulation research is a topic of extensive debate, especially regarding the efficacy of sham control methods for transcranial direct current stimulation (tDCS) studies. Fassi and Cohen Kadosh [1] assessed the influence of subjective participant belief regarding stimulation type (active or sham) and dosage on behaviour using data from Filmer et al. [2] who applied five stimulation protocols (anodal 1.0mA, cathodal 1.0mA, cathodal 1.5mA, cathodal 2.0mA and sham) to assess the neural substrates of mind wandering. Fassi and Cohen Kadosh [1] concluded that subjective belief drove the pattern of results observed by Filmer et al. [2]. Objective: Fassi and Cohen Kadosh [1] did not assess the key contrast between conditions in Filmer et al. (2019), 2mA vs sham, rather they examined all stimulation conditions. Here, we consider the relationship between objective and subjective intervention in this key contrast. Methods: We replicated the analysis and findings of both Filmer et al. [2] and Fassi and Cohen Kadosh [1] before assessing 2mA vs. sham via Bayesian ANOVA on subjective belief regarding stimulation type and dosage. Results: Our results support objective intervention as the strongest predictor of stimulation effects on mind-wandering when 2mA vs sham was examined, over and above that of subjective intervention. Conclusions: The conclusions made by Filmer et al. [2] are confirmed. However, it is important to control for and understand the possible effects of subjective beliefs in sham controlled studies. Best practice to prevent these issues remains the inclusion of active control conditions.

Evaluating the Acceptance of Hemp Food in Australian Adults Using the Theory of Planned Behavior and Structural Equation Modelling

This research presents a mixed methods (qual-QUANT) approach to the evaluation of the intention to consume hemp foods in an Australian sample soon after its legalization, using the Theory of Planned Behavior (TPB). Structural equation modeling was used to evaluate items developed from semi-structured interviews, with a focus on the TPB factors; attitudes toward hemp food consumption, subjective beliefs, and perceptions of control. Findings support the notion that consumers may be confused about associations between Cannabidiol (CBD) oil, tetrahydrocannabinol (THC), and hemp food produced from Cannabis sativa. Highly salient negative associations are mediated by the perception of positive aspects of CBD for some consumers, but the value placed on others’ acceptance of hemp food is the greatest indicator of intention to consume hemp food products. It is suggested that greater education of consumers might allay fears borne of association of hemp food to either CBD or THC, and any move toward disassociation of hemp food to either entity would have positive repercussions for the hemp food industry. Findings have implications for other novel foods that carry highly salient negative associations for consumers.

Accounting for Confirmation Bias in Crowdsourced Label Aggregation

Collecting large-scale human-annotated datasets via crowdsourcing to train and improve automated models is a prominent human-in-the-loop approach to integrate human and machine intelligence. However, together with their unique intelligence, humans also come with their biases and subjective beliefs, which may influence the quality of the annotated data and negatively impact the effectiveness of the human-in-the-loop systems. One of the most common types of cognitive biases that humans are subject to is the confirmation bias, which is people's tendency to favor information that confirms their existing beliefs and values. In this paper, we present an algorithmic approach to infer the correct answers of tasks by aggregating the annotations from multiple crowd workers, while taking workers' various levels of confirmation bias into consideration. Evaluations on real-world crowd annotations show that the proposed bias-aware label aggregation algorithm outperforms baseline methods in accurately inferring the ground-truth labels of different tasks when crowd workers indeed exhibit some degree of confirmation bias. Through simulations on synthetic data, we further identify the conditions when the proposed algorithm has the largest advantages over baseline methods.

Prince: An improved method for measuring incentivized preferences

This paper introduces the Prince incentive system for measuring preferences. Prince combines the tractability of direct matching, allowing for the precise and direct elicitation of indifference values, with the clarity and validity of choice lists. It makes incentive compatibility completely transparent to subjects, avoiding the opaqueness of the Becker-DeGroot-Marschak mechanism. It can be used for adaptive experiments while avoiding any possibility of strategic behavior by subjects. To illustrate Prince’s wide applicability, we investigate preference reversals, the discrepancy between willingness to pay and willingness to accept, and the major components of decision making under uncertainty: utilities, subjective beliefs, and ambiguity attitudes. Prince allows for measuring utility under risk and ambiguity in a tractable and incentive-compatible manner even if expected utility is violated. Our empirical findings support modern behavioral views, e.g., confirming the endowment effect and showing that utility is closer to linear than classically thought. In a comparative study, Prince gives better results than a classical implementation of the random incentive system.

Neural activity tracking identity and confidence in social information

Humans learn about the environment either directly by interacting with it or indirectly by seeking information about it from social sources such as conspecifics. The degree of confidence in the information obtained through either route should determine the impact that it has on adapting and changing behaviour. We examined whether and how behavioural and neural computations differ during non-social learning as opposed to learning from social sources. Trial-wise confidence judgments about non-social and social information sources offered a window into this learning process. Despite matching exactly the statistical features of social and non-social conditions, confidence judgments were more accurate and less changeable when they were made about social as opposed to non-social information sources. In addition to subjective reports of confidence, differences were also apparent in the Bayesian estimates of participants' subjective beliefs. Univariate activity in dorsomedial prefrontal cortex (dmPFC) and posterior temporo-parietal junction (pTPJ) more closely tracked confidence about social as opposed to non-social information sources. In addition, the multivariate patterns of activity in the same areas encoded identities of social information sources compared to non-social information sources.