Humans rely on a powerful mechanism to control their actions – they bind features of their response to stimulus features, which allows for a seamless access of that response upon a future encounter of the same situation. An open question is how these bindings come about: Is the success of a response, the mere co-activation of a stimulus and a response, or the intended goal of the action the driving force of binding? We present an approach that disentangles these three accounts by examining binding effects for action slips, i.e., unintended, erroneous responses. Participants provided speeded responses to letters and we assessed binding of task-relevant stimulus features and response features through sequential analyses of performance as a function of the extent of feature overlap between trials and the success of responding. The results support the view that successful and unsuccessful episodes enter bindings and that these bindings pertain to the intended, correct response rather than to the executed, erroneous response. This finding qualifies binding as an immediate measure to learn from errors.

For the efficient control of goal-orientated behavior, the human cognitive system is provided with several mechanisms regarding stimulus selection, encoding, and processing. One of these mechanisms is the binding of stimuli and responses. According to the theory of event coding (Hommel, 1998, Hommel, 2004), an object and the belonging features of that object are integrated together with the executed response in one event file. Every reencounter with one or more of the stored features leads to the automatic retrieval of the previously constructed event file including the response features, thereby influencing ongoing actions. One factor that is assumed to modulate feature integration (but has not yet been systematically investigated in the area of action control) is the salience of a feature (Hommel, 2005). That is, features might be weighted by perceptual salience, leading to a benefit of salient features or stimuli for the integration process. Furthermore, an influence of salience on binding effects could be assumed for task-relevant target stimuli (in Stimulus-Response Binding), as well as for task-irrelevant distractor stimuli (in Distractor-Response Binding). A first series of experiments (N > 100) suggests that the influence of salience is not as substantial as assumed. The perceptual salience of an irrelevant stimulus does not per se lead to larger binding effects.

In this study we investigated processes of stimulus-response binding and retrieval for erroneous responses. Are mistakes stored and retrieved from memory, like correct responses, leading to a perpetuation of erroneous behavior? Or do people take into account that an error was made, storing and retrieving only correct or corrected responses? Two experiments were designed to assess SR binding and retrieval effects in a sequential prime/probe design using a colour classification task, with an orthogonal variation of response relation (colour repetition vs. change) and stimulus relation (word repetition vs. change). Experiment 1 (n=83) measured SR binding and retrieval for errors that occurred under time pressure (chance errors), while Experiment 2 (n=81) required participants to deliberately respond erroneously (instructed errors) in some of the trials. We found standard SR binding effects following accurate responses in the prime in both experiments. However, no significant SR binding effects were obtained in trials following erroneous prime responses, indicating either that errors are not bound and/or retrieved from memory, or that retrieval of both erroneous and corrected responses might have neutralized each other.

When facing particular combinations of stimuli and responses, people create temporary event files integrating the corresponding stimulus and response features. Repeating one or more of these features retrieve the entire event file, which impairs performance if not all features repeat (partial-repetition costs). We studied how durable event files are over time and how sensitive they are to intervening objects or stimulus-response events. After-effects of relevant and irrelevant stimulus-response bindings were assessed after intervals of 1-5 seconds between creation and retrieval of the binding that were either unfilled (Experiment 1A), filled with 0, 2, or 4 presentations of the same neutral stimulus (1B), or of changing stimuli (1C), or filled with 0, 2, or 4 task-unrelated stimulus-response combinations (2A) or the same number of repetitions of the binding-inducing stimulus-response combination (2B). Taken altogether, the findings show a strong impact on the duration of the interval but no systematic effect of the type and number of intervening events. This suggests that event files disintegrate over time, as a function of spontaneous decay, but not due to interference from other bindings.

Storage and retrieval of episodic stimulus-response (SR) bindings are core mechanisms of action regulation. Yet, it is an unresolved issue how these transient SR bindings relate to longer lasting learning effects. Empirical findings are scarce and unsystematic so far. The present talk explores the relation between transient SR bindings and principles of Pavlovian Conditioning. A series of experiments addresses to which extent SR binding and learning effects reflect similar or different mechanisms. We used an overshadowing procedure to test whether transient binding effects for distractors “mimic” typical overshadowing effects, that is, whether binding effects are attenuated for a distractor presented together with another, but more salient, distractor. This approach has the potential to unravel how processes of selective prioritization will impact on SR binding. Furthermore, it will provide first insight to which extent transient SR bindings might be understood as the “cognitive bases” of longer lasting learning effects.

How do we adapt to shield ourselves against distractors so that these do not prevent us from achieving our goals? This question is usually investigated through conflict tasks as e. g. the Stroop task. In particular, researchers are interested in the sequential congruency effect (SCE), which is characterized by a modulated influence of task irrelevant information after different levels of conflict. Theoretically, there have been two opposing explanations for this effect. While control accounts explain the SCE as top-down attentional adaptation process, binding accounts attribute this effect to the bottom-up binding of stimulus-response (S-R) associations. Recently, the idea was raised that not only S-R associations could be bound (and later retrieved) but also the state of cognitive control itself, i. e. an attentional configuration that is independent of a specific S-R combination. In a first experiment we used a Stroop task in which we provided task-irrelevant context information. Based on previous research, we predict that in sequential trials the repetition of this context information enhances the retrieval of previously bound cognitive control and, therefore, leads to a stronger SCE. Going beyond these results, we wanted to replicate the rapid decay of S-R bindings for the binding of cognitive control. For this reason, we manipulated the retrieval delay speculating that longer delays impair the retrieval of cognitive control, i. e. the context-specific SCE. We will discuss the implications of the results for the integration and extension of control and binding accounts.