Item-specific SA and SC priming
Responding to a stimulus according to the current task demands can entail two processes (Horner & Henson,
2011,
2012; Moutsopoulou et al.,
2015). On the one hand, a stimulus (e.g., car) can require a task-specific semantic classification (e.g., small/large). On the other hand, a stimulus can require an action (e.g., left/right manual response). This then potentially leads to the formation of stimulus–classification (SC) associations (e.g., car–large) and stimulus–action (SA) associations (e.g., car–right; Moutsopoulou et al.,
2015). After such item-specific SC and SA associations have been formed, they can be tested via item-specific repetition priming effects by comparing item-specific switches and repetitions of SC/SA mappings between a stimulus’ prime trial and probe trial (see, e.g., Henson et al.,
2014; Logan,
1988,
1990, for earlier accounts of stimulus–response associations and repetition priming). The retrieval of SC/SA associations leads to faster responses and fewer errors when the item-specific required classification/action repeats compared to when the required classification/action does not match the one implied by the previously established SC/SA association for that stimulus (item-specific SC/SA priming effect; Horner & Henson,
2011,
2012; Moutsopoulou et al.,
2015; Pfeuffer et al.,
2017). More specifically, there is evidence for facilitation after SC/SA repetition (e.g., Horner & Henson,
2011,
2012) as well as for interference when SC/SA mappings switch from prime trial to probe trial (Horner & Henson,
2011,
2012; Moutsopoulou et al.,
2015).
In the item-specific priming paradigm used in the present study, participants were to classify everyday objects by pressing one of two response keys. Everyday objects are displayed only once as a prime and once as a probe with a lag of two to seven trials in between (Moutsopoulou et al.,
2015; Pfeuffer et al.,
2017; Pfeuffer, Hosp et al.,
2018). Item-specific SC and SA mappings are orthogonally repeated/switched between an item’s prime and probe trial, allowing for independent assessment of item-specific SC and SA priming effects in probe trial reaction times and error rates (comparing item-specific switches and repetitions of SC/SA mappings between prime and probe).
SA and SC associations were previously reported to be independent components of stimulus–response (SR) associations (Horner & Henson,
2009; Moutsopoulou & Waszak,
2013; Moutsopoulou et al.,
2015; Pfeuffer et al.,
2017). That means that reaction time (RT) and error rates independently increased for item-specific SC and SA switches (between prime and probe) compared to repetitions (Horner & Henson,
2009; Moutsopoulou & Waszak,
2013; Moutsopoulou et al.,
2015; Pfeuffer et al.,
2017).
Importantly, by assessing item-specific priming effects, Pfeuffer et al. (
2017), Pfeuffer, Hosp et al. (
2018) and Pfeuffer, Moutsopoulou et al. (
2018) found that both, SC and SA associations were formed not only by active response execution (called ‘executed blocks’), but also by passive listening to verbal instructions (called ‘verbally coded blocks’) denoting the relevant classification and the respective action for a given stimulus during a prime trial. As an example, participants viewed the image of an apple and heard the verbal codes ‘small, right’ via headphones. This corresponds with findings from other experimental paradigms consistently showing that item-specific SR associations can be formed following single-trial response execution and following mere single-trial instruction without response execution (Cohen-Kdoshay & Meiran,
2007; Liefooghe & De Houwer,
2018; Liefooghe et al.,
2012; Meiran & Cohen-Kdoshay,
2012; Meiran et al.,
2017; Oshrit & Meiran,
2009; Ruge & Wolfensteller,
2010; Ruge et al.,
2018,
2019; Wenke et al.,
2007).
Expected switch probability based on experience
Previous studies found that expectations based on previous experience modulated SR associations and conflict (Aben et al.,
2017; Abrahamse et al.,
2016; Leboe et al.,
2008). Moreover, context-specific modulation of performance based on previous experiences (though not item-specific priming effects) has been reported (Crump & Logan,
2010; Leboe et al.,
2008; Thomson et al.,
2013,
2014). For instance, Crump and Logan (
2010) showed that priming effects differed depending on location cues. When location was predictive of high or low task switch probability, task switch costs were reduced/increased based on experienced task switch probabilities, respectively. Thus, there is evidence that experienced probabilities influence performance.
Furthermore, experienced probabilities of conflict or task switches can become associated with specific items (Chiu & Egner,
2017; Chiu et al.,
2020; Jacoby et al.,
2003; Leboe et al.,
2008). In these conflict or task switching studies, specific stimuli were presented multiple times with specific proportions of task switches or conflict (incongruent mappings). For stimuli with experienced high probabilities of task switches/conflict, the effects of task switches/conflict decreased compared to stimuli with low probability of task switches/conflict (e.g., item-specific proportion congruency effect: e.g., Bugg et al.,
2008; Jacoby et al.,
2003; Schmidt & Besner,
2008); or item-specific switch probability effect (e.g., Chiu & Egner,
2017; Chiu et al.,
2020; Kang & Chiu,
2021). These effects of experienced probabilities can be explained by associative learning mechanisms. That is, the experienced probability (over multiple trials) of a stimulus being paired with a task switch/conflict leads to adaption processes like contingency learning (Blais & Bunge,
2010; Jacoby et al.,
2003; Schmidt,
2013; Schmidt & Besner,
2008).
Finally, there is evidence that experienced switch probability influences item-specific SC/SA priming effects (Pfeuffer et al.,
2020) even when stimuli are primed and probed only once. Across the prime and probe instances of stimuli, Pfeuffer et al. (
2020) realized different experienced item-specific SC/SA repetition/switch probabilities in different groups of participants (e.g., in Exp. 2: frequent condition: item-specific SC repetition and SA switch > 70%; infrequent conditions: 3 other combinations of item-specific SC and SA repetitions/switches < 30% in total). Most importantly, an item-specific repetition/switch was only experienced once per stimulus in the probe when the effect of this item-specific repetition/switch was tested. Therefore, item-specific switch probabilities were only realized across the entire list of stimuli. That means, item-specific associative learning of classifications/actions or repetition/switch probabilities was impossible. Nevertheless, when SC/SA switches were frequent (across different previously encountered stimuli), the respective item-specific SC/SA priming effects for novel primed and probed stimuli were reduced as compared to when SC/SA repetitions were frequent. That is, there were modulatory effects of switch probability experienced across different previously encountered stimuli on the size of priming effects in probe trials.
As participants were unable to explicitly report repetition/switch proportions afterward, Pfeuffer et al. (
2020) argued that their effects were based on implicit learning processes (see also Blais et al.,
2012; Crump & Logan,
2010; Thomson et al.,
2014). Thus, SC/SA priming effects were modulated based on participants’ expectation based on experiences across different previously encountered stimuli. This modulation of item-specific priming effects by experienced switch probability was similarly observed after both, active response execution and passive listening to verbal codes during the corresponding prime trials (Pfeuffer et al.,
2020).
The present study: expected switch probability based on instruction
Importantly, so far, the influence of expectations on item-specific priming effects (i.e., on the encoding and/or retrieval of corresponding associations) has only been tested based on expectations induced by own previous experience. In contrast, the influence of explicit expectations on SC/SA priming in the absence of expectation-inducing experience (e.g., based on mere instruction) has not been assessed. Specifically, it is unknown whether and how an explicit instruction that item-specific repetitions/switches are frequent, which induces an expectation, affects item-specific SC/SA priming effects. Here, we thus investigated whether item-specific SC/SA priming effects are also modulated by expected switch probabilities induced by mere instruction. Moreover, in contrast to previous studies, we assessed expected switch probabilities that changed frequently and unpredictably across the course of the experiment. Doing so, we aimed to assess whether participants were able to rapidly adapt (i.e., by changing their SC/SA associations and showing reduced/increased SC/SA priming effects) to frequently varying item-specific switch expectations based on instructions (i.e., explicit knowledge).
In other research fields, expected probabilities based on prior experience versus instruction were investigated, for instance, in the context of the description–experience gap in risky decision making (Dutt et al.,
2014; Hau et al.,
2008; Hertwig & Erev,
2009; Park et al.,
2021). In fact, participants seem to treat experienced and instructed probabilities differently leading to contrasting decisions (Hau et al.,
2008). One key difference between description (instruction) and experience is that people tend to give less weight to small probabilities (rare events) when they decide based on experience compared to description (Hertwig & Erev,
2009). Based on these findings, it seems possible that expectations based on instruction (present study) versus based on experience (Pfeuffer et al.,
2020) differently modulate item-specific priming effects.
Additionally, as priming can be described as the facilitation of different processes (e.g., perception, classification, action), it is most interesting that also verbal codes (i.e., instruction) can lead to item-specific priming effects as previously mentioned (Pfeuffer et al.,
2017; Pfeuffer, Moutsopoulou et al.,
2018). While item-specific priming effects based on execution were extensively investigated (e.g., Horner & Henson,
2011,
2012; Moutsopoulou et al.,
2015), it is not quite clear yet how verbal codes lead to the formation of SR associations. Recent studies on priming effects of motor imagery (Liefooghe et al.,
2021; Palenciano,
2021) suggest that motor imagery might be a contributing factor, but, for instance, subvocal rehearsal and visual imagery also cannot be excluded as of now.
On the one hand, previous studies found strong similarities between SC and SA associations formed by active task execution and by verbal coding (Pfeuffer et al.,
2017; Pfeuffer, Hosp et al.,
2018). But on the other hand, priming effects were found to be larger after execution compared to after verbal coding (Pfeuffer et al.,
2017,
2020; Pfeuffer, Hosp et al.,
2018) and, specifically, only the multiple execution but not multiple prime trials of verbal coding of SC/SA mappings led to increased priming effects. (Pfeuffer, Moutsopoulou et al.,
2018). This points towards essential differences between the mechanisms underlying these types of priming. Most importantly for the present study, priming effects based on execution and verbal coding were similarly influenced by expectations based on experienced switch probability (Pfeuffer et al.,
2020). Nevertheless, to rule out that only one type of priming might be susceptible to instructed switch proportions, we additionally compared execution and verbal coding. That is, we were interested in whether processes of SC and SA priming based on execution and verbal coding are equally susceptible to modulation by expectations based on instruction. Alternatively, verbally coded SR associations might be more susceptible to expectations based on instructed switch probability than already executed SR associations simply, for instance, due to a common declarative representational format.
The present study set out to address these questions based on a modified version of the item-specific priming paradigm previously used by Pfeuffer and colleagues.
First, instead of establishing different switch probabilities through actual experience, in the present study item-specific instructed switch probabilities (25% vs. 75%) were induced via cues. Experienced item-specific switch probability remained constant at 50% throughout the experiment. That is, unpredictably, for half of the stimuli the item-specific SC/SA mappings repeated between the stimulis’ prime and probe, whereas they switched for the other half of stimuli. Second, item-specific instructed switch probabilities changed randomly across blocks allowing for within-subject comparisons. These two procedural modifications enabled us to explicitly test the hypothesis that item-specific SC/SA priming effects are modulated by item-specific switch xpectations induced by instruction (i.e., by explicit knowledge) and are flexibly adapted from block to block.
To our knowledge, no study has previously assessed the influence of item-specific switch expectations derived from explicit knowledge (mere instruction) on item-specific priming and SC/SA encoding/retrieval. Nevertheless, our hypothesis that explicit knowledge and resulting expectations can modulate item-specific priming effects is somewhat supported by previous studies regarding modulations of response conflict. These studies suggest that trial-wise instructions might be sufficient to modulate response conflict and to induce expectation-based proportion congruency effects in the Simon task (Desender,
2018; Wühr & Kunde,
2008) and in the Stroop task (Bugg & Smallwood,
2016; Bugg et al.,
2015; Entel et al.,
2014). Importantly, however, in these studies (with the exception of a single experiment in Bugg et al.,
2015), induction blocks were used, where the instructed switch probability was experienced, to ensure that instructions seemed valid to participants (Bugg & Smallwood,
2016; Bugg et al.,
2015; Entel et al.,
2014). However, by confounding instructed and experienced switch probability and testing stimuli multiple times, these studies could not differentiate between influences of experience and mere instruction (i.e., explicit knowledge). As mentioned above, this confound was avoided in the present study, as instructed item-specific switch probability (25% or 75%) changed randomly across prime-probe blocks, while the experienced item-specific switch probability was held constant at 50%. Therefore, we were able to investigate the modulation of item-specific SA and SC priming effects (i.e., corresponding SC/SA associations) based on expected switch probabilities induced by mere instruction which were not strengthened by experienced switch probabilities.
To summarize, the present study addressed two research questions concerning the modulation of SC and SA priming effects by expectation:
If SC and SA associations were encoded and retrieved solely based on mnemonic processes, SC and SA priming effects should remain unaffected by merely expected (instructed), but never experienced item-specific SC/SA switch probabilities. Vice versa, a modulation of SC/SA priming effects by merely instructed item-specific SC/SA switch probabilities would imply that participants adapted to expectations induced by instruction/explicit knowledge. We hypothesize that item-specific priming effects can be modulated on the basis of not only previous experience (Pfeuffer et al.,
2020), but also expectations derived from mere instruction and will be stronger under instructed 25% switch probability, compared to instructed 75% switch probability.
We additionally assessed if SC/SA associations formed by different types of prime trials (executed vs. verbally coded) were similarly modulated by instructed switch probability (interactions of prime type, instructed switch probability, and action/classification). On the one hand, one could expect that both prime types are similarly affected by expectation, as Pfeuffer et al. (
2020) found no difference between these prime types regarding the influence of expectation based on experienced switch probability. On the other hand, differences between the prime types (Pfeuffer et al.,
2017,
2020; Pfeuffer, Hosp et al.,
2018) could influence the susceptibility to instructed switch probability. Thus, observing similar or different effects of instructed switch probability depending on prime types would also be informative regarding the mechanisms of priming (facilitation of perception, classification or action) which can be modulated by expectations derived from instruction/explicit knowledge.