Feelings of responsibility and temporal binding: A comparison of two measures of the sense of agency.

Temporal binding refers to a subjective shortening of the interval between an action and its perceptual consequences. Temporal binding has often been used by researchers to indirectly measure participants' sense of agency (SoA), or the subjective sense of causing something to happen. Other studies have proposed links between temporal binding and feelings of moral responsibility. The present study compared subjective interval estimates to feelings of responsibility in a between-subjects design (Exp 1) and a within-subjects design (Exp 2). Participants either estimated the interval between two events (two tones in the passive condition, or a keypress followed by a tone in active conditions) or rated their feeling of responsibility for the tone(s). Manipulations of participant involvement and choice impacted feelings of responsibility more than temporal estimates. Overall, the two dependent variables followed different patterns, suggesting subjective interval estimates may not be a reliable proxy for feelings of responsibility.

When dyads act in parallel, a sense of agency for the auditory consequences depends on the order of the actions.

The sense of agency (SA) is the perception of willfully causing something to happen. Wegner and Wheatley (1999) proposed three prerequisites for SA: temporal contiguity between an action and its effect, congruence between predicted and observed effects, and exclusivity (absence of competing causal explanations). We investigated how temporal contiguity, congruence, and the order of two human agents' actions influenced SA on a task where participants rated feelings of self-agency for producing a tone. SA decreased when tone onsets were delayed, supporting contiguity as important, but the order of the agents' actions (lead, follow, or simultaneous) also mattered. Relative contiguity was the main determinant of SA, as delayed tones were usually attributed to the most recent action. This was unaffected by contingencies between the two actors' actions (Experiment 2), showing that contiguity has a powerful influence on SA, even during joint action in the presence of other cues.

Predictable and self-initiated visual motion is judged to be slower than computer generated motion.

Self-initiated action effects are often perceived as less intense than identical but externally generated stimuli. It is thought that forward models within the sensorimotor system pre-activate cortical representations of predicted action effects, reducing perceptual sensitivity and attenuating neural responses. As self-agency and predictability are seldom manipulated simultaneously in behavioral experiments, it is unclear if self-other differences depend on predictable action effect contingencies, or if both self- and externally generated stimuli are modulated similarly by predictability. We factorially combined variation in (1) predictability of action effects, (2) spatial congruence, and (3) performance by the self or computer to dissociate these influences on a visual discrimination task. Participants performed 2AFC speed judgments. Self-initiated motion was judged to be slower than computer-initiated motion when action effect contingencies were predictable, while spatial congruence influenced speed judgments only when action effect contingencies were unpredictable. Results are discussed in relation to current theories of sensory attenuation.

Cognitive load decreases the sense of agency during continuous action.

The sense of agency that normally accompanies voluntary actions depends on a combination of sensory predictions and other inferences. For example, when people manipulate moving objects and rate their degree of control, control ratings are influenced by proximal correlations between motor commands and visual feedback as well as the overall success or failure of the action. The relative importance of sensory predictions vs. post hoc feedback may depend on the availability and perceived reliability of those cues, which is context dependent. The present study investigated how increasing cognitive load during a visuomotor tracking task influences the sense of agency, and whether cognitive load influences the extent to which control ratings depend on sensory predictions vs. post hoc feedback. In Experiments 1 and 2, participants performed a dual task that involved tracking a moving target using a joystick while rehearsing 3, 5, or 7 digits. Control ratings decreased as memory set size increased, even though set size had no significant effect on objective tracking error. Experiment 3 replicated this finding while also manipulating the favorability of feedback presented after each trial. Control ratings were correlated with post hoc feedback, but there was no significant interaction between feedback and cognitive load. These results suggest that sensorimotor predictions, performance feedback, and availability of working memory resources can all influence sense of agency. The hypothesis that people rely more on post hoc feedback to rate control when they are distracted was not supported.

Relationship between BOLD amplitude and pattern classification of orientation-selective activity in the human visual cortex.

Orientation-selective responses can be decoded from fMRI activity patterns in the human visual cortex, using multivariate pattern analysis (MVPA). To what extent do these feature-selective activity patterns depend on the strength and quality of the sensory input, and might the reliability of these activity patterns be predicted by the gross amplitude of the stimulus-driven BOLD response? Observers viewed oriented gratings that varied in luminance contrast (4, 20 or 100%) or spatial frequency (0.25, 1.0 or 4.0 cpd). As predicted, activity patterns in early visual areas led to better discrimination of orientations presented at high than low contrast, with greater effects of contrast found in area V1 than in V3. A second experiment revealed generally better decoding of orientations at low or moderate as compared to high spatial frequencies. Interestingly however, V1 exhibited a relative advantage at discriminating high spatial frequency orientations, consistent with the finer scale of representation in the primary visual cortex. In both experiments, the reliability of these orientation-selective activity patterns was well predicted by the average BOLD amplitude in each region of interest, as indicated by correlation analyses, as well as decoding applied to a simple model of voxel responses to simulated orientation columns. Moreover, individual differences in decoding accuracy could be predicted by the signal-to-noise ratio of an individual's BOLD response. Our results indicate that decoding accuracy can be well predicted by incorporating the amplitude of the BOLD response into simple simulation models of cortical selectivity; such models could prove useful in future applications of fMRI pattern classification.

Is that what I wanted to do? Cued vocalizations influence the phenomenology of controlling a moving object.

The phenomenology of controlled action depends on comparisons between predicted and actually perceived sensory feedback called action-effects. We investigated if intervening task-irrelevant but semantically related information influences monitoring processes that give rise to a sense of control. Participants judged whether a moving box "obeyed" or "disobeyed" their own arrow keystrokes (Experiments 1 and 2) or visual cues representing the computer's choices (Experiment 3). During 1s delays between keystrokes/cues and box movements, participants vocalized directions ("up", "down", "left", or "right") cued by letters inside the box. Congruency of cued vocalizations was manipulated relative to previously selected keystrokes and upcoming box movements. In Experiment 1, reported obey moves and feelings of control reflected the true frequency of obey moves, but were also modulated by vocalizations. Incongruent vocalizations reduced reported obey moves, whereas congruent vocalizations increased them. In Experiment 2, vocalizations had stronger effects when their congruence with primary-task box movement was consistent for several consecutive moves before congruence changed. In Experiment 3, analogous impacts of vocalizations occurred when the computer selected the directions and participants judged whether the computer had control of the box. We conclude that predicted and perceived action-effects associated with semantically related but separate and ostensibly irrelevant actions can be conflated with one another. This interference is not restricted to actions performed with the same effector or within the same modality, or even by the same actor. Thus in estimating degrees of control, the mind integrates across ongoing action systems, whether or not they are logically task-relevant.

Taking credit for success: the phenomenology of control in a goal-directed task.

We studied how people determine when they are in control of objects. In a computer task, participants moved a virtual boat towards a goal using a joystick to investigate how subjective control is shaped by (1) correspondence between motor actions and the visual consequences of those actions, and (2) attainment of higher-level goals. In Experiment 1, random discrepancies from joystick input (noise) decreased judgments of control (JoCs), but discrepancies that brought the boat closer to the goal and increased success (the autopilot) increased JoCs. In Experiment 2, participants raced to the goal against a computer-controlled rival boat while varying levels of noise interfered with each boat. Participants reached the goal more often and rated their own control higher when the computer rival had good control. Subjective control over moving objects depends partly on consistency between motor actions and their effects, but is also modulated by perceived success and competition.

Do implicit and explicit measures of the sense of agency measure the same thing?

The sense of agency (SoA) refers to perceived causality of the self, i.e. the feeling of causing something to happen. The SoA has been probed using a variety of explicit and implicit measures. Explicit measures include rating scales and questionnaires. Implicit measures, which include sensory attenuation and temporal binding, use perceptual differences between self- and externally generated stimuli as measures of the SoA. In the present study, we investigated whether the different measures tap into the same self-attribution processes by determining whether individual differences on implicit and explicit measures of SoA are correlated. Participants performed tasks in which they triggered tones via key presses (operant condition) or passively listened to tones triggered by a computer (observational condition). We replicated previously reported effects of sensory attenuation and temporal binding. Surprisingly the two implicit measures of SoA were not significantly correlated with each other, nor did they correlate with the explicit measures of SoA. Our results suggest that some explicit and implicit measures of the SoA may tap into different processes.

The phenomenology of controlling a moving object with another person.

The phenomenology of controlling what one perceives is influenced by a combination of sensory predictions and inferential processes. While it is known that external perturbations can reduce the sense of control over action effects, there have been few studies investigating the impact of intentional co-actors on the sense of control. In three experiments, we investigated how individuals' judgments of control (JoC) over a moving object were influenced by sharing control with a second person. Participants used joysticks to keep a cursor centered on a moving target either alone or with a co-actor. When both participants' actions had similar perceptual consequences, JoC ratings were highest when self-generated movements were the only influence on the cursor, while the appearance of sharing control with a second person decreased JoC ratings. By contrast, when participants performed complementary actions with perceptually distinctive consequences, JoC ratings were highest when both participants were able to influence the cursor. The phenomenology of control during joint action is influenced by low-level visuomotor correlations, the presence of competing causal influences, and group-level performance.