Sixteen weeks of regular, brief meditation practice significantly changed neuronal activity related to executive control functions in the Stroop task. These changes were, however, not accompanied by related improvements in behavioral performance and did not pertain to the late negative ERP component (400–600 ms) that typically reflects the behavioral interference effect in the Stroop task (e.g., Liotti et al., 2000
; Hanslmayr et al., 2008
Meditation practice led to a relative increase of lateral posterior N2 amplitudes over both hemispheres, irrespective of stimulus congruency. Estimation of the neural sources (VARETA) suggests that these changes in the meditation group were primarily driven by increased activity in the left medial and lateral occipitotemporal areas for congruent stimuli, contrasted by decreased activity in similar brain areas in the control group. These left-hemispheric areas of the ventral processing stream have previously been identified as being selectively involved in lexical tasks (e.g., Cohen et al., 2002
; Cohen and Dehaene, 2004
; Shaywitz et al., 2004
), with a similar posterior N2 component as observed here (e.g., Adorni and Proverbio, 2009
). It thus seems plausible that this effect reflects more successful or consistent attentional amplification, selective to the word stimuli that were used in this task. This interpretation is in line with the time course of enhanced stimulus processing when attending to non-spatial features of a stimulus. Typically, enhanced negative posterior ERP amplitudes appear from around 100 to 150 ms after stimulus onset (Hillyard and Anllo-Vento, 1998
; Hillyard et al., 1998
). Even more, the posterior N2 is particularly enlarged when attending to the color as compared to the form of a stimulus (Eimer, 1997
). Thus, while the control group exhibited a habituation effect over the course of the study (and 3 × 144 trials), which was expressed in a reduction of the ERP amplitudes and the related cortical source strengths, the meditation group showed the opposite pattern, where increased activation of task relevant cortical areas developed with meditation practice.
The second difference between meditators and controls was observed in the P3 component (310–380 ms). The majority of ERP studies of the Stroop task focus on later components starting around 400 ms, as these tend to correlate with behavioral performance (Liotti et al., 2000
), whereas the preceding P3 component appears to reflect earlier aspects of stimulus processing that, in themselves, are not the source of the behavioral Stroop interference effect (Ilan and Polich, 1999
). Changes of the P3 over the course of the study were primarily observed for incongruent stimuli. While the participants in the control group exhibited an increase of the P3 amplitude for incongruent stimuli, a decrease was observed for the meditation group. The P3 decrease in electrode space was accompanied by significantly decreased signal strength in source space, which comprised lateral occipitotemporal and inferior temporal regions of the right hemisphere. These areas have been implicated in object recognition processes (Schendan and Kutas, 2002
; Schendan and Stern, 2007
). In addition, the temporal/parietal P3 component is considered to reflect attentional resource activation that is generated when perceptual stimulus discrimination occurs and is linked to related inhibition processes that are required when conflicting stimulus information is present (Polich, 2007
). The pattern of results emerging for the P3 component thus suggests that through meditation practice the perceptual processing of incongruent stimuli becomes less resource demanding.
These findings bear similarities to the results from a previous study, where experienced meditators showed a reduced P3 amplitude to a distracter tone during an auditory oddball stimulation while they were meditating (Cahn and Polich, 2009
). There are however, noteworthy differences to our study. In Cahn and Polich's study a meditation state was compared to a neutral thinking state, whereas we studied the effect of meditation in a task that was performed outside of the meditation practice. Furthermore, we investigated changes through meditation practice developing over time, while Cahn and Polich, (2009
) only tested at one time point and thus do not directly address the question of causal influences of meditation training. The parallels are nevertheless interesting, as they suggest that an ability that developed and is present during meditation practice appears to generalize to a different task performed when not meditating. It may indicate that state effects observed during meditation may translate into trait effects observed outside of meditation (Cahn and Polich, 2006
), an assumption that underlies the idea that meditation practice generalizes into daily activities and extends to contexts separate from meditation practice (Hodgins and Adair, 2010
; Slagter et al., 2011
Furthermore, our results are in line with other studies suggesting that meditation practice leads to more effective brain resource allocation (Slagter et al., 2007
). Slagter and co-workers employed the attentional blink paradigm to investigate how a three-month intensive meditation retreat changes the temporal deployment of attention compared to a non-meditating matched control group (Slagter et al., 2007
). During the attentional blink task participants have to attend to a rapidly changing stream of stimuli (e.g., letters) and report the identity of two target stimuli (e.g., digits) after each trial. Performance to the second target in the stream is typically negatively affected if it appears within 500 ms after the first target, the so-called attentional blink effect (Shapiro et al., 1997
). After the meditation retreat the meditators showed a reduced attentional blink effect. Furthermore, the P3b amplitude elicited by the first target stimulus was reduced in meditators after the retreat and the participants with the greatest decrease of the P3b amplitude also showed the largest decrease in attentional blink size (Slagter et al., 2007
). Interestingly, the additional analysis of the phase of oscillatory theta activity following successfully detected second targets, showed a reduced cross-trial variability, considered to indicate that the deployment of attention was more consistent and that through meditation training attentional resources become more rapidly available to process additional information (Slagter et al., 2009
The results from a recent fMRI study comparing meditators and matched controls on the Stroop task provide further support for our findings. Compared to a control group, meditators showed reduced activity in various brain areas subserving attention (Kozasa et al., 2012
). The authors interpret their overall pattern of findings as evidence of enhanced efficiency in meditators that may result from improved sustained attention and impulse control.
When considering our results of enhanced N2 and decreased P3 amplitudes and source strengths in light of the reviewed findings, a possible interpretation emerges. We surmise that the more successful attentional amplification of the color word stimuli evidenced by increased N2 amplitudes/source strengths had the subsequent effect that fewer resources needed to be invested during object recognition processes, especially when incongruent stimulus information was processed, indexed by the decrease in P3 amplitudes/source strengths.
Confining the meditation training to a very simple, but fundamental, mindful breathing meditation, which often constitutes the first step into a more elaborate path of different meditation practices, gives confidence that the observed changes indeed stem from the meditation practice itself. Having kept the group sessions to a bare minimum (a total of 3 h), makes it furthermore unlikely that unspecific group effects account for the changes. The fact that participants only meditated for very brief periods each day speaks against an explanation that life style changes could explain the observed differences, an influence that may well be relevant when studying the effects of longer daily meditation practices, of meditation retreats or when studying highly experienced meditators.
As meditation effects were compared to effects in a non-active waitlist control group, an alternative explanation might be that the observed effects merely result from the fact that the meditators were engaged in a novel regular activity per se, rather than being specific to the meditation practice. The current design cannot fully rule this out, but given that the observed effects are in line with results from several other studies into similar meditation practices, it appears likely that the effects are more specific. However, the general weakness of waitlist controlled designs in this respect needs to be acknowledged. The study tells us that engaging in 10 min of daily meditation practice for the given period has specific effects. It can, however, not be concluded that these effects are completely unique to meditation practice in general or to this specific type of mindfulness meditation in particular. While the mindfulness training had these effects, other practices or activities may have as well. Future studies will have to face up to the challenge of addressing the question how specific changes associated with meditation training actually are. Toward this end, control conditions that are matched with respect to somatic, mental, and cognitive demands but without actually being meditation practice will be required.
In this study the participants were required to record frequency and amount of meditation practice themselves. As the experimenters appeared to have a good rapport with the participants and it was emphasized that it is more important to provide accurate information than to fulfill a specific regime, we have no specific reason to doubt the honesty and accuracy of these records. We are, however, in no position to objectively confirm this. The fact that we found a positive relationship between mindfulness (FFMQ) and amount of meditation practice might be taken as a positive indicator, but as both are self-report measures they may be prone to similar distortions. Future studies may want to control actual meditation time more objectively. One needs to be aware, though, that this is only possible to a certain extent, because even if, for example, actigraphic measures of rest and activity cycles were available or sensors were integrated into meditation stools or cushions, we have to rely on participant reports whether during a period of physical rest they actually engaged in meditation practice.
An unexpected result of the study was that no differences in behavioral measures between meditation and control group appeared. This finding goes hand in hand with the lack of an effect of meditation practice on the LN, but is at odds with results from several other studies, which tended to show better performance of meditators over controls in similar measures of executive attention and conflict resolution (Chan and Woollacott, 2007
; Jha et al., 2007
; Moore and Malinowski, 2009
). One important difference between such cross-sectional data and the study presented here is that a longitudinal design requires the repeated administration of the same experimental task. In the current study 144 trials of the Stroop task were administered at each time point (to a total of 432 trials). The fact that overall RTs did not improve after T2 (T2: 632 ms, T3 632 ms) and that accuracy was above 95% for incongruent trials, suggests that a performance ceiling might have been reached. A further difference to the cross-sectional study that showed the clearest performance differences between mindfulness meditators and a control group (Moore and Malinowski, 2009
) was, that a verbal paper-pencil version of the Stroop task was used, whereas here a computerized version with manual button presses was employed. Several authors have highlighted that the way of administering the Stroop task has an influence on behavioral results and the interference effects in particular (Kindt et al., 1996
; Salo et al., 2001
). Liotti and co-workers (2000) furthermore showed that different response formats in the Stroop task (verbal, covert, or button press responses) yield differential scalp distributions of the ERPs. Variations in task administration, trial repetition, and related ceiling effects or the type and duration of the investigated meditation practice may have contributed to some diversity in outcomes observed in different studies (Chiesa et al., 2011
An additional explanation is suggested by new evidence regarding the involvement of the anterior cingulate cortex (ACC) in Stroop-like tasks. The ACC has been shown to be the generator of the LN and to be involved in performance monitoring and response selection (Liotti et al., 2000
; Hanslmayr et al., 2008
). However, two recent event-related fMRI studies suggest that the role of the ACC is more related to anticipatory regulation of attention rather than the specific selection of responses itself (Roelofs et al., 2006
; Aarts et al., 2008
). The lack of differential effects in the LN might thus reflect that with extended exposure to the Stroop task anticipatory regulation was perfected in both groups, resulting in the observed ceiling effect. The meditation practice, it seems, has improved earlier stages of processing (indexed by N2 and P3 changes) that reflect more fundamental changes in attentional processing and are less amenable to simple task repetition effects. Although speculative, this would also explain why clear behavioral differences are found when meditators encounter the Stroop tasks for the first time (Chan and Woollacott, 2007
; Jha et al., 2007
; Moore and Malinowski, 2009
), while they tend not to develop on repeated presentation of the same task as was observed here and also reported before (Anderson et al., 2007
Our results also appear at odds with findings from a longitudinal study carried out by Lutz and co-workers, who found a reduction in RT variability (Lutz et al., 2009
) that was not present in out data. There are, however, noteworthy differences to our study in that Lutz et al. investigated changes after much more intensive meditation training (a three-month retreat) and studied the response to rare targets in an auditory task. It might well be that a combination of the already mentioned ceiling effect and the considerable difference in the amount of training accounts for the different outcome.
Despite the lacking evidence of behavioral effects of the meditation practice, significant differences on self-reported mindfulness levels were evident and the increase in mindfulness (FFMQ-total) was correlated with the amount of time participants invested in their meditation practice, suggesting that the time invested in meditation directly translates into recognizable increases in mindfulness. Considering the sample size of N = 12 for this analysis, one needs to be cautious, though, to not over-interpret the results of this correlation.
This study focused on the effects of meditation practice on mechanisms of attentional control as indexed by performance and ERP measures related to the Stroop task. However, we do assume that also other aspects of attention may have been influenced by the meditation practice. A recent paper provides an excellent theoretical account, arguing that mindfulness meditation training, developed over longer periods of time, should lead to the enhancement of cognitive core processes including the sustained monitoring of one's own mental states, the ability to disengage from distracting objects and the skill to redirect attention back to the chosen focus (Slagter et al., 2011
). We suggest that the observed changes in the N2 and P3 partially reflect the enhancement of such core processes. In line with this view of more wide-ranging changes, our study also included various other measures, results of which we aim to report elsewhere. These pertain to sustained attention and alertness and the orienting of attention without interfering or conflicting stimuli. In addition, these data will allow us to investigate brain dynamics during rest and meditation practice, where we are particularly interested in global brain states, indexed by oscillating neural activity. Several recent studies suggest that there might be differences between meditators and non-meditators (e.g., Lutz et al., 2004
; Tei et al., 2009
; Cahn et al., 2010
) and between different types of meditation (Travis and Shear, 2010
) in this respect. Although not directly related to the methodological approach we were using, it is also worth noting that several studies comparing meditators and non-meditators found differences in brain structure (cortical thickness or gray matter), often in brain areas involved in attentional functions (Lazar et al., 2005
; Hölzel et al., 2008
; Luders et al., 2009
; Grant et al., 2010
) and first longitudinal studies show such structural changes in gray and white matter even after relatively brief periods of meditation practice (Tang et al., 2010
; Hölzel et al., 2011