Bachelor Thesis Brain and Cognition
Unconscious Cognitive Control and Metacognition
Amke Müller University of Amsterdam
Amke Müller
Supervisor: Filip van Opstal Word counts: 5036
Abstract
Cognitive control enables us to adapt our behaviour to the ever-changing environment. Experimentally, cognitive control mechanisms have been investigated by conflict adap-tation paradigms, with the Gratton effect as behavioural expression. In the current study, we investigated the role of metacognitive experience and unconscious information in conflict adaptation, using a metacontrast paradigm. With different inter-trial intervals, we manipulated the effects of unconscious priming and metacognition, since uncon-scious priming effects decay quickly. Replicating earlier findings, we found a Gratton
effect, irrespectively of the length of the inter-trial interval and the visibility of the
prime. This implicates that metacognitive experience of the conflict, rather than uncon-scious priming effects, is the underlying mechanism inducing conflict adaptation. These findings question the existence of unconscious cognitive control and support the view that subjective experience is the crucial factor to insert top-down control of behaviour.
The Role of Metacognitive Awareness in Unconscious Cognitive Control
One of the most interesting features of human cognition is its extreme flexibility in reaction to external stimuli and the adaptability to the rapidly changing environment (Botvinick, Braver, Barch, Carter & Cohen, 2001). Accountable for this accomplishment is cognitive control, which includes the ability to regulate, organise, evaluate, and correct thoughts and actions according to our current goals (Braver, 2012). Generally, cognitive control is thought to be uniquely to humans, guided by consciousness (Desender, Van Lierde & Van den Bussche, 2013). The logic behind this strong link between cognitive control and consciousness is the idea that we are normally aware of changes that need adaptation of our behaviour (Kunde, Reuss & Kiesel, 2012). This has been grounded in the Global Neuronal Workspace Theory, which proposes that consciousness is obliga-tory to induce cognitive control (Dehaene & Naccache, 2001). According to this theory, uncon-scious information is unable to activate top-down control since it does not have the qualities re-quired to access to multiple resources, such as reciprocal, synchronised, long-lasting, widespread activation (Dehaene & Naccache, 2001). Following this line of reasoning, consciousness can be seen as hallmark for cognitive control.
Experimentally, the relationship between consciousness and cognitive control has been ex-tensively studied with conflict adaptation paradigms. Conflict adaptation, as a behavioural signature of cognitive control, can be defined as adjustments of behaviour to improve performance after a conflict is detected (Desender et al., 2013). Typically, conflict arises when task-relevant infor-mation interferes with task-irrelevant inforinfor-mation. A common paradigm to test conflict adaptation is the Eriksen flanker task (Eriksen & Eriksen, 1974). In this task, participants view a target which is flanked by a stimulus that either calls for the same response (i.e. congruent) or the opposite re-sponse (i.e. incongruent). This target can be, for example, an arrow pointing to the left or right with a flanker arrow pointing either to the same or opposite direction. Participants are asked to respond to the target arrow (i.e. relevant information) and to ignore the flanker arrow (i.e.
(RTs) are faster and error rates are lower on congruent than on incongruent trials (Lamers & Roe-lofs, 2011). Interestingly, this congruence effect is sensitive to the previous trial: it is smaller after an incongruent trial than after a congruent trial. An explanation for this reduction of the congruence
effect is that incongruent trials trigger a conflict between task-relevant and task-irrelevant
infor-mation. Consequently, the need of control for task-irrelevant information is enhanced, leading to longer RTs and higher error rates. This observation, known as the Gratton effect, is commonly seen as a reliable measurement of cognitive control (Desender et al., 2013; Gratton, Coles and Donchin, 1992).
To this date, the relationship between the Gratton effect and consciousness remains highly debated with controversial results (for an overview Kunde, Reuss & Kiesel, 2012). Several authors have claimed that consciousness of the conflict is necessary to trigger the Gratton effect (Green-wald, Draine & Abrams, 1996; Kunde, 2003; Kunde & Wühr, 2006). For example, Kunde (2003) investigated the influence of consciousness on the Gratton effect with a meta-contrast paradigm. Participants were asked to respond to a target arrow, either pointing leftward or rightward. These target arrows were preceded by a prime arrow that exactly fitted in the contour of the target, there-by, the target acted as a mask (Kunde, 2003). Furthermore, prime visibility was manipulated by ap-plying different duration times of prime presentation, resulting in one condition with a perceptually visible prime and one condition with a masked prime (Kunde, 2003). Interestingly, the results showed that the reduction in the prime-target congruence effect after an incongruent trial only takes place in the visible prime condition. This finding implicates that the Gratton effect is only evident when the subjects are conscious of the response conflict (Kunde, 2003). Consequently, conflict ad-aptation is not a direct consequence of response conflict, but rather an intention-mediated strategy (Kunde, 2003). At the same time, other studies observed adaptation effects also when subjects were unaware of the conflict, providing evidence for unconscious conflict adaptation (Van Gaal, Lamme & Ridderinkhof, 2010; Desender et al., 2013). The most compelling evidence for unconscious con-flict adaptation was provided by Van Gaal and colleagues (2010), using a similar meta-contrast
par-adigm to Kunde. Importantly, they applied some slight variations, such as, a shortening of the inter-trial interval and the elimination of the warning signal in order to assure that participants stay alert during the whole task (Van Gaal et al., 2010). With these adjustments, Van Gaal et al. (2010) ob-served conflict adaptation effects not just after conscious conflict but also after unconscious con-flict. Given these results, they refuted the idea that intention-mediated strategies are necessary to trigger conflict adaptation (Van Gaal et al., 2010). However, it can be questioned whether these ef-fects are truly unconscious or rather a conscious by-product (Desender & Van den Bussche, 2012).
Another possibility is that the subjective experience of conflict, rather than the conflict it-self, induces conflict adaptation. This subjective experience, also known as metacognition, is a form of introspection and generally associated with conscious access (Fleming, Dolan & Frith, 2012; Fletcher & Carruthers, 2012). This hypothesis can be grounded in the extended conflict monitoring theory. In this framework, the Anterior Cingulate Cortex (ACC) appears to play a crucial role in conflict adaptation (Botvinick, 2007). Particularly, the ACC is involved in conflict monitoring as well as in adjustments of top-down control. Regarding the meta-contrast paradigm, it has been ob-served that the ACC is more active during incongruent than during congruent trials, which can be interpreted as a need of more control on incongruent trials (Botvinick, 2007). Furthermore, recent evidence suggests that the ACC also evaluates action outcomes to guide decision making, by its in-volvement in avoidance learning (Botvinick, 2007). It has been observed that the ACC responds to the aversive aspect of the conflict, triggering avoidance behaviour (Gehring & Willoughby, 2002; Frank, 2005). Overall, these findings provide considerable evidence for the role of emotions in cog-nitive control. Moreover, metacognition and emotion are robustly linked to each other, as metacog-nition plays an important role in emotion regulation (Fernandez-Duque, Baird & Posner, 2000). To put it differently, this could explain the role of metacognitive experience, namely, metacognition could act as the aversive signal to adapt behaviour (Questienne, Van Opstal, Van Dijk, & Gevers, 2016).
Although this hypothesis is still speculative, recently, the role of metacognition in conflict adaptation has been experimentally demonstrated (Desender, Van Lierde & Van den Bussche, 2014).Desender and colleagues (2014) used a similar meta-contrast paradigm as previous studies, but the crucial difference was that they provided a measurement of metacognition. For this purpose, participants had to respond to either leftward or rightward pointing target arrows that were preceded by prime arrows (Desender et al., 2014). These primes were either congruent or incongruent, thus, eliciting the congruency effect. Crucially, metacognition was measured by asking participants after every trial whether they experienced conflict or not (Desender et al., 2014). This made it possible to differentiate whether the actual conflict or the subjective experience of the conflict led to conflict adaptation. Interestingly, their results suggest that the Gratton effect is in fact dependent on the metacognitive experience of conflict and independent of the presence of an actual conflict (Desender et al., 2014). In line with the extended conflict monitoring theory, this study provides compelling evidence for the role of metacognition in conflict adaptation (Desender et al., 2014). However, this evidence for the role of metacognition in conflict adaptation cannot be considered as conclusive. One criticism about their study, is the way of assessing metacognitive experiences. By asking participants directly about their metacognitive experience this measurement is per definition subjective and, in addition, susceptible to potentially biases (Questienne et al., 2016).
For this reason, our study applies a different strategy to measure metacognitive in a more objective manner. Based on the fact, that effects of metacognitive experiences and unconscious priming differ in their duration, it is possible to exclude the effect of unconscious priming and measure the effects of metacognitive experiences. For instance, unconscious priming effects decay quickly and are usually absent when the inter-trial interval is longer than 500 ms (Dehaene,
Naccache, Cohen, Bihan & Mangin, 2001; Greenwald et al., 1996). However, some studies showed unconscious priming effects up to 2 seconds after stimulus onset (Cohen, Van Gaal, Ridderinkhof, Van den Wildenberg & Lamme, 2009). On the other hand, the onset of metacognitive experiences can be timed to 300-400 ms after the stimulus is presented, since metacognition is associated with
the P3-component (Desender, Van Opstal, Hughes & Van den Bussche, 2016). Furthermore, con-scious information (i.e. metacognitive experiences) can be held on-line for longer periods and can be stored in the working memory (Shimamura, 2000). Given these differences in durations, here we applied two inter-trial intervals to manipulate metacognitive experiences and unconscious effects: a long inter-trial interval condition (duration 3.5 seconds), which ensures that unconscious priming effects are completely decayed before the next trial starts and a short inter-trial condition with an inter-trial interval of 500 ms. Within the short inter-trial-interval both metacognition and uncon-scious priming effects could be accountable for the adaptation effect. This leads to the expectation that, if metacognition is the crucial factor to induce conflict adaptation, both conditions show the same adaptation effects. Whereas, if unconscious information triggers conflict adaptation, it is ex-pected that the adaptation effect is only present in the short inter-trial interval condition. The para-digm to examine unconscious adaptation conflict will be a similar meta-contrast parapara-digm that also has been used in previous research to measure conflict adaptation (Desender et al., 2013; Kunde, 2003; Van Gaal et al., 2010). With these adjustments, this study aims to bring clarification to the debate about unconscious and metacognitive influences on cognitive control.
Methods
Participants
In total twenty-eight students, 15 male and 13 female, were recruited via lab.uva.nl at the University of Amsterdam. For their participation, they got one course credit (for sixty minutes). Mean age of the sample was 22 years (SD=2.1, range 20-29). All participants reported normal or corrected-to-normal vision. The study was approved by the ethics committee, and the participants were provided with a written informed consent. Because of technical problems with the metacon-trast task, the data of two participants was excluded from the analysis. Another participant was re-moved from further analysis, because his reaction time differed more than two SD from the mean reaction time. Regarding the prime detection task, data was collected from nineteen of the twenty-eight students.
Stimuli and apparatus
Stimuli were presented using a PC running Presentation software (Psychtoolbox) on a 60 Hz monitor. Primes and targets were arrows that pointed to the left or right. The stimuli were presented in black on a light-grey background. For half of the trials, the mask stimuli were a small replica of the target arrow and exactly fitted into the contour of the target arrow (meta-contrast mask), while for the other trials the primes did not fit into the contour of the target arrow (pseudo-mask; see fig-ure 1). This design enables us to manipulate prime visibility, with the meta-contrast mask masking the prime and the pseudo-mask not masking the prime. Responses were accessed using a standard QWERTY keyboard.
Material
Meta-contrast Paradigm
Conflict adaptation was measured by a meta-contrast task. Participants were instructed to indicate the direction of the target arrows as accurately and quickly as possible. They were asked to press the ‘d’ key with their left index finger in response to a leftward-pointing arrow, and to press the ‘k’ key in response to a rightward-pointing arrow. Each trial started with a fixation cross (dura-tion 1,000 ms), next the prime arrow was flashed briefly (dura(dura-tion 16.66 ms), followed by a blank screen (duration 33 ms), and after that the target arrow appeared (duration 150 ms). On half of the trials the target arrow acted as a mask for the prime arrow since the prime arrow exactly fitted into the contour of the target arrow (meta-contrast mask condition), on the other half the prime did not fit into the contour and therefore not masked the target (pseudo-mask condition). Before the exper-iment started participants practiced on 24 trials. After this, the main experexper-iment started which com-prised eight blocks, containing 80 trials each (resulting in a total of 640 trials), that were separated by a short break. Half of the trials had a long inter-stimulus interval with a duration of 3,500 ms
(long-ITI condition) and the other half had a short inter-stimulus interval with a duration of 500 ms (short-ITI condition). Further, on half of the trials, the prime and the target pointed in the same di-rection (congruent) and the other half they pointed in the opposite didi-rection (incongruent). The tri-als were presented in a random order each block. Additionally, we counterbalanced the two inter-trial intervals by randomly assigning subjects either to the condition with the short inter-stimulus interval first, followed by the long inter-stimulus interval or vice versa.
Prime detection task
Prime visibility was assessed by the prime detection task. Participants were presented with the same stimuli as during the meta-contrast task, but this time they had to identify the direction of the prime arrow rather than the direction of the target arrow. Moreover, feedback about the perfor-mance was provided after every trial by the colour of the fixation cross (green for correct, red for incorrect). In this task accuracy was important, not the reaction time. The prime detection task con-sisted eight blocks, each containing 60 trials (resulting in a total of 480 trials). Again, participants were randomly distributed over the two conditions with either the short inter-stimulus interval first, followed by the long inter-stimulus interval or vice versa.
Fig.1. Experimental design. (A) Metacontrast task. A leftward or rightward pointing arrow, is
fol-lowed by a pseudo-mask (A) or metacontrast mask (B). After the onset of the mask, participants had to indicate the direction of the target, while ignoring the prime. The inter-trial interval (ITI) var-ied from 500 ms for the short-ITI condition and 3,500 ms for the long-ITI condition. (B) Prime de-tection task. The prime dede-tection task was similar to the metacontrast task: the same stimuli have been used, with the difference that participants were ask to decide on the orientation of the prime, while ignoring the mask. Another difference was that the mask was not followed by a blank inter-val, but rather by another fixation cross that indicated whether the response was right (green) or wrong (red).
Procedure
Participants were invited to the laboratory at the University of Amsterdam. After signing the informed consent, participants were randomly distributed over the two conditions: participants per-formed either the long ITI first, followed by the short ITI or they perper-formed the short ITI first, fol-lowed by the long ITI. After the instruction, the participants first completed the meta-contrast task (duration 30 minutes). Next, prime visibility was assessed by asking participants to perform the
prime detection task (duration 30 minutes). The same conditions used in the meta-contrast task
ap-plied to the prime detection task. The duration of the tasks altogether was about sixty minutes.
Expectations
To explore the role of metacognition in conflict adaptation, we are interested in the
congru-ence effect (incongruent trials – congruent trials) of the different conditions. Our focus is the Grat-ton effect, thus the reduction of the congruence effect, expressed in longer RTs and error rates, after
a previous incongruent trial. In our study, we applied three manipulations resulting in a 2x (ITI-short or ITI long) 2x (pseudo-mask or meta-contrast mask on trial n-1) 2x (congruent or
incongru-ent trial n-1) design. For our hypothesis, we are mainly interested in differences in the size of the
congruence effect, between the ITI conditions, after an incongruent trial (n-1), with a masked prime
(n-1, i.e. unconscious). As mentioned earlier, our hypothesis is that metacognition, rather than un-conscious priming effects, is accountable for conflict adaptation after an unun-conscious conflict. Thereby, we expect metacognition to be present in both conditions of the inter-trial interval manipu-lation (short-ITI and long-ITI), embodied by the Gratton effect. Alternatively, if unconscious prim-ing effects cause conflict adaptation after an unconscious conflict, it is expected that the Gratton
effect appears only when the ITI is short and not when the ITI is long. We expect to display these
effects in both outcome measures: RTs and error rates.
Regarding prime visibility, it is expected that participants in the meta-contrast condition do not perform better than by chance. For the pseudo-mask condition, it is expected that participants detect the primes better than by chance.
Data analysis
Meta-contrast paradigm
The relation between metacognition and the Gratton effect was investigated by comparing the size of the congruence effect in the long inter-trial interval condition with the short inter-trial condition. A repeated measures ANOVA examined the size of the congruence effect (mean RT
in-congruent – mean RT in-congruent of the current trial). Thereby, only correct trials are taken into
analysis with within-subjects’ variables: prime-target congruence on previous trial n-1 (congruent vs. incongruent), inter-trial interval (short vs. long) and conscious or unconscious prime (meta-contrast mask vs. pseudo-mask). The outcome measure is the size of the congruence effect dis-played in error rates and reaction time. The first trial of each block (1.25%) was removed from fur-ther analysis. Moreover, for the RTs, all trials following errors (12.5%) and RTs above 1,000 ms and under 100 ms, were excluded from the analysis.
Prime-detection task
To ensure that primes were invisible, we used the data from the detection task to compute a signal detection d’ (‘d-prime’). This measure is obtained by treating one level of the response cate-gory as signal (i.e., ‘left-pointing prime arrow’) and the other level as noise (i.e., ‘right-pointing prime arrow’). The response ‘left’ is considered as a ‘hit’ (H) when the arrow points leftward and as a ‘false alarm’ (F) when the arrow points rightward. To characterize the performance of the sub-jects, the d’ is calculated which is a measure of the difference between H and F. Therefore, the z-transforms of the H and F rates are calculated. This makes it possible to compare measures with dif-ferent ranges of absolute value, and takes the variability of difdif-ferent measures into account. d’ is then calculated by subtracting the transformed false alarm rate (z(F)) from the transformed hit rate (z(H)). A d’ of 0 shows that participants were not able to differentiate between signal and noise. By using a paired sample t-test it is analysed, if the d’ value significantly differs from chance. It is ex-pected that the d’ value not significantly differs from chance level for the masked prime, while the opposite is expected for the pseudo mask.
Results
Conflict adaptation reflected in RTs
Mean sizes of the congruence effect, reflected in reaction times, of all factorial combinations of the included variables are presented in Tabel 1 (a). The mean sizes of the congruence effect of correct trials were entered into a repeated measures ANOVA with as within-subject variables: length of inter-trial interval (2 levels: ITI short or long), mask of the previous trial (2 levels: meta-contrast mask or pseudo-mask) and congruence on the previous trial (2 levels: congruent or incon-gruent).
This analysis showed a main effect of inter-trial interval (F(1,23) = 5.142, p = 0.033, 𝜼𝜼²=.183). The congruence effect was smaller after a short inter-trial interval (42 ms) than after a long inter-trial interval (52 ms). Furthermore, there was no significant main effect of the mask
(F(1,23) = 1.283, p = 0.269). However, contradicting to our expectations, there was no main effect of congruence of the previous trial, F < 1, which indicates the absence of a Gratton effect. No sig-nificant interactions have been found. Taken together, these findings indicate that conflict adapta-tion was not reflected in RTs.
Table 1
Mean Size of Congruence Effect Reflected in (a) Reaction Times (in ms) and (b) Error Rates (in percentages) as a Function of Masking (Meta-Contrast Mask and Pseudo-Mask) in Trial n-1 and Prime-Target Congruence in Trial n-1 (incongruent and congruent) for the Different Inter-Trial Interval Conditions (a) Incongruent Congruent Inter-trial interval M SD M SD Unconscious Short (500 ms) 45 29 39 23 Long (3,500 ms) 50 24 48 14 Conscious Short (500 ms) 47 24 39 43 Long (3,500 ms) 48 19 62 28 (b) Incongruent Congruent Inter-trial interval M SD M SD Unconscious Short (500 ms) 45 29 39 23 Long (3,500 ms) 50 24 48 14 Conscious Short (500 ms) 47 24 39 43
Long (3,500 ms) 48 19 62 28
Conflict adaptation reflected in error rates (ER)
When conducting the same repeated measures ANOVA on the size of the correspondence effect in error rates, a different pattern was found. Mean size of correspondence effect in error rates, for all factorial combinations of the included variables, are listed in Table 1(b).
A significant main effect of previous congruence was observed, (F(1,23) = 4.8, p = 0.039, η²
=.173). The congruence effect was smaller if the previous trial was incongruent (7.7%), than if the previous trial was congruent (10%). This reduction in the congruence effect after a previous incon-gruent trial reflects the Gratton effect. There was also a significant main effect of inter-trial interval, (F(1,23) = 28.8, p<.001, η² =.556). The size of the congruence effect was smaller in the long inter-trial condition (3.6%), compared with the short inter-inter-trial condition (14.1%). Again, there was no significant main effect of mask, F < 1. Furthermore, there was no interaction effect between previ-ous mask and previprevi-ous congruence, indicating that the Gratton effect was equally present in both conditions (pseudo-mask and meta-contrast mask).
Crucially, there was no significant interaction between previous congruence and inter-trial interval, F < 1 (See Figure 1). These findings implicate that the Gratton effect was present inde-pendent of the duration of the inter-trial interval. No other interaction effects have been found. Tak-en together, this result implicates that after response conflict adaptation takes place irrespectively of the masking, or the inter-trial interval.
Figure 1. Mean size of the congruence effect reflected in error rates, measured with the
meta-contrast paradigm, as a function of previous congruence and inter-trial interval. Error bars represent a 95% confidence interval.
Prime visibility
To assess prime detection, we calculated d’, as explained above. Subjects with hit or false alarm probabilities more than 2 SD above the average were excluded, leaving 18 subjects for the data analysis. For the meta-contrast condition, the 18 subjects correctly identified 60% of the
primes (SD = 12.89). The mean d’ value was 0.65, which differed significantly from chance, t(17) = 3.811, p = 0.001. For the pseudo-mask condition 63% of the primes were correctly identified (SD = 15.80). The mean d’ value was 0.75 and also differed significantly from chance, t(17) = 3.656, p = 0.002. These results imply that the participants in both conditions perceived the primes better than by chance. Although this effect was expected to be true for the conscious condition, we did expect participants to perform not better than by chance in the unconscious condition. To put it another way, these results implicate that participants in both conditions perceived the primes consciously.
Given that these findings imply that our manipulation of prime visibility failed, conclusions must be interpreted with caution.
Discussion
In the current study, we investigated the role of metacognition in conflict adaptation. There-fore, we used a masked prime paradigm to examine conflict adaptation, with two different inter-trial intervals manipulating unconscious priming effects and metacognition. We found reliable conflict adaptation in both the short inter-trial interval and the long inter-trial condition, suggesting that metacognition, rather than unconscious priming effects, is accountable for conflict adaptation. Error rates clearly showed a reduction in congruence effects following an incongruent trial (n-1), irrespec-tively of the mask and the inter-trial interval. Consequently, the Gratton effect can be ascribed to the metacognitive experience of the conflict. These findings are contradicting to the study from Van Gaal and colleagues (2010), where they claim for the existence of unconscious cognitive control. Hence, we demonstrated that an unconscious conflict can trigger conflict adaptation, not just imme-diately after the conflict, but also after unconscious priming effects already decayed. These findings indicate, that conflict adaptation is rather driven by metacognitive experience than by unconscious information. Moreover, these apparently contradicting findings could be mere due to different in-terpretations instead of different results. Technically, Van Gaal and colleagues (2010) merely found that unconscious conflict is leading to conflict adaptation and interpreted these results as unscious cognitive control. Although these results indeed suggest that conunsciously perceiving a con-flict is not necessary to induce cognitive control, nothing can be said about the mechanisms under-lying unconscious conflict adaptation. Other factors, such as, the subjective feeling of the conflict could also have been the cause for conflict adaption. This is exactly what our study demonstrated and what also has been implicated by Desender and colleagues (2014). Altogether it can be con-cluded that metacognition is the crucial component to induce conflict adaptation, independent of the prime visibility.
While our study showed reliable Gratton effects in error rates, reaction times did not reveal conflict adaptation, which differs from previous studies about unconscious conflict adaptation (Desender et al., 2014; Kunde et al., 2013; Questienne et al., 2016; Van Gaal et al., 2010). An ex-planation for these different findings can be a common observation known as the speed/accuracy tradeoff, which describes the relationship between reaction times and error rates in performing a task. For example, while performing a task subjects can either concentrate on reaction times while making more errors or the other way around. This phenomenon has also been demonstrated with different conflict adaptation paradigms (Ullsperger, Bylsma & Botvinick, 2005; Francken et al., 2011). As an illustration, Francken and colleagues (2011) showed, while using the same meta-contrast task, notably higher effects of conflict adaptation in error rates than in reaction times. Their conclusion was that this reduction in conflict adaptation reflected in reaction time could be due the high error rate, in line with the speed/accuracy tradeoff (Francken et al., 2011). Since we found even higher error rates after a conflict (≈17%), than Francken et al. (2011) found in their study (≈11%), the speed/accuracy tradeoff in our study could be even stronger. As a result, conflict adap-tation is only reflected in error rates and not in reaction times. Nonetheless, to bring further clarifi-cation to the expression of conflict adaptation in a meta-contrast paradigm, future work is needed to investigate the relationship between error rates and reaction times. For example, a comparison be-tween two conditions one with the instruction to respond as quickly as possible and the other to re-spond as accurately as possible, could clarify the exact relationship of error rates and reaction times in this context.
After all, it should be noted that the conclusions we have drawn are not without limitations. A possible concern could be the manipulation of prime visibility. The results of the prime detection task indicate that subjects were aware of the primes in both conditions (meta-contrast mask and pseudo-mask). Consequently, one could argue that our study did not examine unconscious induced conflict adaptation, but only conscious induced conflict adaptation. However, previous studies, us-ing the same meta-contrast paradigm, found reliably differences in the perception of the primes with
masked primes being detected not better than by chance (Francken et al., 2011). Similarly, reliable differences have been found between unconscious and conscious triggered conflict adaptation, when using a similar meta-contrast paradigm (Desender et al., 2014; Kunde, 2003; Van Gaal et al., 2010). This makes it implausible that the manipulation of prime visibility in the meta-contrast para-digm failed. Alternatively, it is possible that the prime detection task did not measure prime visibil-ity accurately. A confounding factor here could be the feedback provided about accuracy of the prime detection. While in previous studies feedback was either given after each block or no feed-back at all, here we provided subjects with feedfeed-back after every trial (Francken et al., 2011). Con-sequently, subjects could have focused on the feedback rather than on the prime itself, thereby looked for other indicators of the prime orientation than the prime itself. Moreover, previous re-search has shown that priming detection tasks are influenced by many factors, such as the nature of the target, attentional factors and the delay between stimulus presentation and response (Vermeiren & Cleeremans, 2012). Taken together, these findings imply that the prime detection task is not a very robust measurement of prime visibility. A better approach could be to include not only objec-tive measures, like the prime detection task, but also subjecobjec-tive measures of prime visibility. Previ-ous research suggests that only if both measurements are included, it can be concluded that task per-formance is guided by conscious knowledge (Pasquali, Timmermans & Cleermans, 2010). Future research should examine whether objective and subjective measurements of prime visibility in this context measure the same concept and how this is related to task performance on the prime detec-tion task.
Finally, some suggestions should be mentioned for future research about the measurement of metacognitive experiences. In our study, we manipulated unconscious effects and metacognitive experiences by varying the inter-trial interval. As mentioned earlier, this is a more objective way to measure metacognitive experience than, for example, by self-report (see page 2). Important to real-ise is that this measurement assumes metacognitive thinking to be present before 500 ms after the stimulus onset and to be remained until 3,500 ms after stimulus onset. This assumption is based on
the fact, that EEG measurements have linked metacognitive thinking to the P3 component, which appears 300-400 ms after stimulus onset (Desender et al., 2016). Although this might be true, neu-ronal measurements are required ensure that participants in this study design were truly engaging in metacognitive thinking. With these measurements, it would be possible to differentiate metacogni-tive experiences from unconscious effects on a neuronal basis. Future work is necessary to further specify the temporal dynamics of metacognitive thinking in conflict adaptation paradigms.
In sum, we conclude that metacognition rather than unconscious priming effects induce con-flict adaptation, confirming Desender et al. (2014). Metacognition could be the underlying mecha-nism of unconscious adaptation effects, indicating that adaptation effects cannot be exerted uncon-sciously. Our study questions the plausibility of unconscious top-down control, by showing that top-down control is dependent on subjective experience. These results further expand the potential role of subjective experiences in cognitive control mechanisms.
References
Botvinick, M. M. (2007). Conflict monitoring and decision making: Reconciling two perspectives on anterior cingulate function. Cognitive, Affective, & Behavioural
Neuroscience, 7(4), 356-366. doi:10.3758/CABN.7.4.356
Botvinick, M. M., Braver, T. S., Barca, D. M., Carter, C. S., & Cohen, J. D. (2001). Conflict monitoring and cognitive control. Psychological review, 108(3), 624. doi:10.1037/0033- 295X.108.3.624
Braver, T. S. (2012). The variable nature of cognitive control: A dual mechanisms framework.
Trends in cognitive science, 16(2), 106-113. doi:10.1016/j.tics.2011.12.010
Cohen, M. X., Van Gaal, S., Ridderinkhof, K. R., & Lamme, V. A. F. (2009). Unconscious er- rors enhance prefrontal-occipital oscillatory synchrony. Frontiers in Human Neurosci-
ence, 3, 54. doi:10.3389/neuro.09.054.2009
Dehaene, S., & Naccache, L. (2001). Towards a cognitive science of consciousness: Basic evidence and a workspace framework. Cognition, 79(1), 1-37. doi:10.1016/S0010- 0277(00)00123-2
Dehaene, S., Nacchache, L., Cohen, L., Bihan, D., L., Mangin, J. F., Poline, J. B., et al. (2001). Cerebral mechanisms of word masking and unconscious repetition priming. Nature
Neuroscience, 4, 752-758. doi:10.1038/89551
Desender, K., & Van den Bussche, E. (2012). The magnitude of priming effects is not inde- pendent of prime awareness. Reply to Francken, Van Gaal & de Lange (2011).
Con-scious cognition, 21(3), 1571-1572. doi:10.1016/j.concog.2012.01.017
Desender, K., Van Lierde, E., & Van den Bussche, E. (2013). Comparing conscious and unconscious conflict adaptation. PLoS One, 8(2), e55976. doi:10.1371/journal.pone. 0055976
role of conflict experience in adaption. Psychological Science, 25(3), 675-683. doi: 10.1177/0956797613511468
Desender, K., Van Opstal, F., Hughes, G., & Van den Bussche, E. (2016). The temporal dy- namics of metacognition: Dissociating task-related activity from later metacognitive processes. Neuropsychologia, 82, 54-64. doi:10.1016/j.neuropsychologia.2016.01.003 Eriksen, B. A., & Eriksen C. W. (1974). Effects of noise letters upon the identification of a tar-
get letter in a nonsearch task. Perception & Psychophysics, 16(1), 143-149. doi:10.3758 BF03203267
Fernandez-Duque, D., Baird, J. A., & Posner, M. I. (2000). Executive attention and metacogni- tion regulation. Consciousness and cognition, 9(2), 288-307. doi:10.1006/ccog.2000. 0447
Fleming, L., Dolan, R. J., & Froth, C. D. (2012). Metacognition: Computation, biology and function. Philosophical Transactions of the Royal Society B, 367, 1280-1286. doi: 10.1098/rtsb.2012.002
Fletcher, L., & Carruthers, P. (2012). Metacognition and reasoning. Philosophical Transactions
of the Royal Society B,367(1594), 1366-1378. doi:10.1098/rstb.2011.0413
Francken, J. C., Van Gaal, S., & de Lange, F. P. (2011). Immediate and long-term priming effects are independent of prime awareness. Consciousness and Cognition, 20, 1793-1800. doi:10.1016/j.concog.2011.04.005
Frank, M. J. (2005). Dynamic dopamine modulation in the basal ganglia: A neurocomputa- tional account of cognitive deficits in medicated and nonmedicated Parkinsonism.
Jour-nal of Cognitive Neuroscience, 17, 51-72. doi:10.1162/0898929052880093
Gehring, W. J., & Willoughby, A. R. (2002). The medial frontal cortex and the rapid pro-cessing of monetary gains and losses. Science, 295, 2279-2282. doi:10.1126/science.1066893
control of activation of responses. Journal of Experimental Psychology: General, 121, 480-506. doi:10.1037/0096-3445.121.4.480
Greenwald, A. G., Draine S. C., & Abrams, R. L. (1996). Three cognitive markers of
uncon-scious semantic activation. Science, 273, 1699-1702.
Doi:10.1126/science.273.5282.1699
Kunde, W. (2003). Sequential modulations of stimulus-response correspondence effects depend on awareness of response conflict. Psychonomic Bulletin & Review, 10(1), 198-205. doi:10.3758/BF03196485
Kunde, W., Reuss, H., & Kiesel, A. (2012). Consciousness and cognitive control. Advances in
cognitive psychology, 8(1), 9-18. doi:10.2478/v10053-008-0097-x
Kunde, W., & Wühr, P. (2006). Sequential modulations of correspondence effects across spa- tial dimensions and taks. Memory & Cognition, 34(2), 356-367. doi:10.3758/BF03193413
Lamers, M. J. & Roelofs, A. (2011). Attentional control adjustments in Erickson and Stroop task performance can be independent of response conflict. The Quarterly Journal of
Experimental Psychology, 64(6), 1056-1081. doi:10.1080/17470218.2010.523792
Pasquali, A., Timmermans, B., & Cleeremans, A. (2010). Know thyself: Metacognitive net- works and measures of consciousness. Cognition, 117, 182-190. doi:10.1016/j.cognition.2010.08.010
Questienne, L., Van Opstal, F., Van Dijk, J. P., & Gevers, W. (2016). Metacognition and cogni- tive control: Behavioural adaptation requires conflict experience. The Quarterly Journal
of Experimental Psychology, 1-15. doi:10.1080/17470218.2016.1251473
Shimamura, A. P. (2000). Toward a cognitive neuroscience. Consciousness and Cognition, 9, 313-323. doi:10.1006/ccog.2000.0450
Ullsperger, M., Bylsma, L. M. & Botvinick, M. M. (2005). The conflict adaptation effect: It’s not just priming. Cognitive, Affective, & Behavioral Neuroscience, 5(4), 467-472. doi:
10.3758/CABN.5.4.467
Van Gaal, S., Lamme, V. A. F., & Ridderinkhof, K. R. (2010). Unconsciously triggered con- flict adaptation. PLoS One 5(7), e11508. doi:10.1371/journal.pone.0011508
Vermeiren, A., & Cleeremans, A. (2012). The validity of d’ measures. PLoS One, 7(2), e31595. doi:10.1371/journal.pone.0031595
