Investigating mechanisms underlying the context-specific proportion congruent effect in a prospective memory paradigm

The project consists of an experimental study that is being conducted as part of a master’s thesis. The aim of the master’s thesis is to shed light on the mechanisms of context-specific adjustments found in so-called conflict tasks that are often used to investigate influences of different experimental manipulations on cognitive control processes. A key finding is that the performance in these conflict tasks is more error prone and slower in conflicting compared to non-conflicting trials (i.e., congruency effect; e.g., Stroop, 1935). The congruency effect is thus supposed to be indicative of how much cognitive control was recruited to resolve conflict between competing stimulus dimensions (e.g., Braem et al., 2019). Various experimental studies manipulating the proportion of conflicting trials in two different contexts (i.e., task-irrelevant contextual features, such as the location on the screen) have now shown that the congruency effect is reduced in the high-demand context, featuring mostly conflicting trials compared to the low-demand context, featuring mostly non-conflicting trials (e.g., Crump, Gong, & Milliken, 2006; Crump et al., 2017; Fischer et al., 2014; Schouppe et al., 2014). This effect is termed context-specific proportion congruent effect (CSPC). Similar effects can also be found on different levels, such as at the level of specific items (i.e., the item-specific proportion congruent effect (ISPC), Bugg, Jacoby, & Chanani, 2011; Shedden et al., 2013) or entire lists of trials (i.e., the list-wide proportion congruent effect (LWPC), Lindsay & Jacoby, 1994; Logan & Zbrodoff, 1979). Initially, these findings were interpreted in terms of attentional control. It was presumed that, over time, the high demand context itself would come to be associated with an attentional control state that is almost activated in an “automatic fashion”, and results in increased attention directed toward task-relevant information as opposed to task-irrelevant information, thereby reducing interference from task-irrelevant information (e.g., Bugg & Crump, 2012; Crump & Milliken, 2009; Fischer et al., 2014). In recent years, however, alternative theories have been put forward to explain this effect. The fundamental premise underlying all of these alternative explanations is the assumption that the CSPC effect does not emerge from higher-order cognitive functions. Instead, it is posited that this phenomenon can be ascribed to either task-inherent factors or low-level learning processes, such as the learning of stimulus-stimulus or stimulus-response associations (i.e., contingency learning, Schmidt, 2016) or formation of temporal expectancies for when a response must be provided (i.e., temporal learning, Schmidt et al., 2014). There are several actions that can be taken when designing experimental paradigms to reduce the influence of stimulus-stimulus or stimulus-response associations (for a review, see Braem et al., 2019). Firstly, Braem et al. (2019) emphasize the usage of different sets of items, so called inducer and diagnostic items. Inducer items refer to items that are utilized to implement the proportion congruent (PC) manipulation. Whereas, diagnostic or transfer items refer to frequency-unbiased items that are used to measure the effects of the PC on performance. It could be shown that CSPC effects are indeed transferred to frequency-unbiased items (e.g., Bejjani et al., 2018; Crump & Milliken, 2009, but see Hutcheon & Spieler, 2017). Secondly, the number of unique stimulus combinations can be increased, up to trial-unique paradigms in which each unique stimulus is presented only once (e.g., King et al., 2012). Thirdly, it has been demonstrated that associations between a demand context and a control setting established in a learning phase can be transferred to different stimulus-response sets as well as entire task sets in a transfer phase (Surrey et al., 2017). It has proven to be much more difficult to develop suitable paradigms that are effective at excluding temporal learning as the sole explanation for the emergence of the LWPC or the CSPC effect (Sprengel et al. 2022). However, using reanalysis of data sets from list-wide proportion congruent paradigms, Cohen-Shikora et al. (2019) were able to show that evidence for the temporal learning account might partially be refuted by using appropriate statistical methods which, for instance, do not rely on the non-linear transformation of RT data, such as general linearized mixed effect models. Yet, even if specifically designed paradigms and appropriate statistical models are employed in studies investigating the CSPC effect, the underlying mechanisms driving the observed effects remain largely unclear. With the aim of investigating the mechanisms underlying the list-wide proportion congruent effect, Bugg et al. (2011) developed a novel design within the classical color-word Stroop task (Stroop, 1935) that incorporated both neutral trials as well as a secondary prospective memory task that required focusing attention on the task-irrelevant stimulus dimension (i.e., the word). Event-based prospective memory (PM) refers to the ability to remember to perform delayed intentions later in the future in response to a specific environmental cue (Einstein & McDaniel, 1990). They conducted two experiments. Experiment 1 demonstrated that the color naming performance in neutral trials (that did not exhibit a frequency bias) was facilitated in the mostly incongruent list where the process of word reading is hypothesized to be attenuated, thereby reducing interference from the irrelevant word. Experiment 2 showed that the weakened word reading in the mostly incongruent list led to impaired performance in the secondary prospective memory task that required participants to detect a specific neutral word. Complementary, the conflict-triggered goal shielding hypothesis (Goschke & Dreisbach, 2008) suggests that when task demands are high (i.e., there is high conflict), attentional resources are fully allocated to goal relevant information. Two different PM tasks were embedded in a spatial compatibility task, featuring mostly incongruent trials. PM task trials could be congruent or incongruent as well. Crucially, one PM task required focusing attention on the stimulus-dimension that was also relevant to the on-going spatial compatibility task (i.e., focal PM task, Einstein & McDaniel, 2005), while the other required focusing attention on the stimulus-dimension that had to be ignored in order to complete the on-going task (i.e., non-focal PM task, Einstein & McDaniel, 2005). The authors assumed that the high level of conflict would lead to shielding of the goal (i.e., the implementation of the spatial compatibility task) at expense of the detection of environmental cues (in that case, the PM cues). In line with this reasoning, they were able to show that non-focal PM cues were overlooked more often compared to focal-PM cues, PM performance was overall impaired by conflict (i.e., incongruent PM cues were overlooked more often), and that this effect was even more pronounced for non-focal compared to focal PM cues. Furthermore, performance in the on-going spatial compatibility task was also impaired by the occurrence of PM cues. In blocks featuring non-focal PM cues compared to those featuring focal PM cues, performance in the on-going task was slowed. The congruency effect was also more pronounced in blocks featuring non-focal PM cues. However, it remains unclear whether the effects described above would be less pronounced in the case of low task demands (i.e., a lower proportion of conflicts) and whether these findings can be transferred to CSPC paradigms in which the demand context changes on a trial-to-trial basis. To address this issue, in this study, we have aimed to develop a CSPC paradigm in which the influence of contingency learning is minimized by (1) increasing the number of unique stimulus combinations and (2) by introducing diagnostic items in the form of rare PM cues that are embedded in the task. The PM cues will only be introduced in a transfer phase following a learning phase in which the CSPC effect is to be established. In addition, the paradigm should be constructed in such a manner that it is not only capable of excluding alternative explanations, but also capable of answering the main research question of whether attentional control plays a role in the emergence of the CSPC effect. To this end, we have introduced two different PM tasks that have to be completed alongside the on-going main experimental conflict task. Crucially, the successful detection of PM cues in one PM task requires attention to be directed to the task-relevant dimension (focal PM task), whereas the other requires attention to be directed specifically to the task-irrelevant dimension that has to be ignored to complete the on-going main experimental task (non-focal PM task). If the high-demand context, over time, is indeed associated with an attentional control state that facilitates the effective selective filtering of irrelevant information in favor of task relevant information, then we would expect the detection of non-focal PM cues to be even more impaired in the high demand context, particularly so if these PM cues are incongruent as opposed to congruent.