- Email: [email protected]
Links between autonomic activity and implicit learning
Links between autonomic activity and implicit learning Alissa B. Forman-Alberti, J. Benjamin Hinnant PII: DOI: Reference:
S0167-8760(16)30738-3 doi: 10.1016/j.ijpsycho.2016.10.014 INTPSY 11191
To appear in:
International Journal of Psychophysiology
Received date: Revised date: Accepted date:
30 June 2016 10 October 2016 17 October 2016
Please cite this article as: Forman-Alberti, Alissa B., Hinnant, J. Benjamin, Links between autonomic activity and implicit learning, International Journal of Psychophysiology (2016), doi: 10.1016/j.ijpsycho.2016.10.014
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
NU
SC
RI
PT
Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
MA
Links between Autonomic Activity and Implicit Learning
Alissa B. Forman-Albertia and J. Benjamin Hinnantb Catholic University of America
D
a
AC CE P
TE
b
Auburn University
Author Note
Alissa B. Forman-Alberti, Department of Psychology, The Catholic University of America; J. Benjamin Hinnant, Department of Human Development and Family Studies, Auburn University Address correspondence to Alissa B. Forman-Alberti, Department of Psychology, 255 O’Boyle Hall, The Catholic University of America, 620 Michigan Ave NE, Washington, DC 20064. Email: [email protected]
1
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
2
Abstract The somatic marker hypothesis posits that autonomic activity occurring in response to specific
PT
stimuli aids in implicit learning, the learning of information without explicit awareness of what
RI
has been learned. This study investigated whether respiratory sinus arrhythmia (RSA), a measure of autonomic nervous system activity, predicted changes in implicit learning. The interaction of
SC
resting RSA and RSA reactivity (change in RSA during the implicit learning task) was
NU
associated with changes in implicit learning, with those who had higher resting RSA and greater RSA withdrawal during the task performing better. These findings contribute to a better
MA
understanding of the autonomic processes that may underlie implicit learning and are discussed in relation to potential links between autonomic activity, implicit learning, and decision making.
AC CE P
making
TE
D
Keywords: autonomic activity; respiratory sinus arrhythmia; implicit learning; decision
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
3
Links between Autonomic Activity and Implicit Learning We learn even when we are not aware of learning taking place. Implicit learning is the
PT
process of gaining sensitivity to novel relationships or regularities among events in the
RI
environment, which are often probabilistic in nature, without conscious knowledge of what has been learned or that learning has even occurred (Reber, 1989). Implicit learning permits
SC
individuals to adapt their behavior to environments without expending conscious effort (Frensch
NU
& Rünger, 2003) and may be experienced as intuitions or “gut feelings” about the available choices (Lo & Repin, 2002). In other words, even though we may not be consciously aware of
MA
learning, our bodies may be subtly influencing our responses to patterns in environmental stimuli.
TE
D
According to the somatic marker hypothesis (SMH; Bechara & Damasio, 2005; Damasio, 1996), autonomic activity plays a key role in implicit learning processes. The SMH suggests that
AC CE P
physiological “markers” occur within individuals in response to encounters with stimuli and that with repeated exposure, these markers then become associated with those events. Somatic markers will be reactivated in response to those stimuli in future experiences, even without conscious awareness of the stimuli’s significance, and will bias cognitive processing and behavior involving them (for a review, see Dunn, Dalgleish, & Lawrence, 2006). Existing research supports the claim that somatic signals, at least initially, assist with learning in uncertain environments, but the majority of this research has focused only on activity in the sympathetic branch of the autonomic nervous system, and has largely ignored the parasympathetic branch. The relationship between the brain and the autonomic nervous system is crucial to understanding the connection between the latter and individuals’ behavior. Both cortical and subcortical areas of the brain contribute to processes such as attentional regulation, affective
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
4
information processing, and behavioral flexibility (Thayer & Lane, 2000). The coordinated activation of all of these processes is what allows individuals to adapt their behavior in the face
PT
of environmental challenges or changes (Thayer, Ahs, Fredrikson, Sollers, & Wager, 2012). For
RI
example, implicit learning involves a functional network of cortical-subcortical connections, with the prefrontal cortex acting as a hub that links brain regions related to memory and learning,
SC
such as the caudate nucleus (Yang & Li, 2012). In order to adapt behavior properly, there needs
NU
to be an appropriate balance between the sympathetic and parasympathetic nervous system (SNS and PNS, respectively); the former stimulates the body during stressful situations when a fight-
MA
or-flight response is needed and the latter promotes readiness to engage with (e.g., learn from) the environment. Research has suggested that this ability to regulate autonomic activity may be
TE
D
related to psychological self-regulation, which also relies on the prefrontal cortex to exert topdown control on other brain regions (Porges, 1992). Given this, PNS activity may be particularly
functioning.
AC CE P
important to consider when examining individual differences in higher-order cognitive
The current study focused on PNS activity and its role as a potential somatic marker in implicit learning. Although this is the first study on this topic, there is evidence that the PNS plays an important role in social and more explicit learning processes (Porges, 2007). To measure PNS activity, the impact of the vagus nerve, one of the primary conduits for parasympathetic influence on multiple organ systems, can be non-invasively assessed. Output from the brain, the nature of which is dependent on an individual’s internal and external circumstances, travels to the heart via the vagus nerve. The vagus nerve acts like a “brake” on the heart (Porges, 2007), increasing cardiac output to facilitate blood flow for active problem solving or promoting fight-or-flight response (i.e., SNS activation) when the brake is withdrawn.
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
5
The ability to flexibly engage and withdraw parasympathetic influence depending on internal and external demands is indicative of adaptive attentional and regulatory processes (Porges, 2003).
PT
This interplay of sympathetic and parasympathetic influence on the heart leads to variability in
RI
heart rate (i.e., heart rate variability or HRV). Because vagal activity fluctuates in phase with respiration (Frazier, Strauss, & Steinhauer, 2004), researchers can document the variability in
SC
inter-beat intervals (i.e., the time between heartbeats) that occurs during typical breathing cycles,
NU
also known as high-frequency HRV (Berntson et al., 1997). This high-frequency HRV is also known as respiratory sinus arrhythmia (RSA). Therefore, RSA acts as an indirect measure of
MA
vagal-parasympathetic influence on the heart.
There is extensive research linking both RSA at rest as well as changes in RSA in
TE
D
response to a stressor (i.e., RSA reactivity) to a host of cognitive outcomes including explicit learning (Fairclough, Venables, & Tattersall, 2005; Haley, Grunau, Oberlander, & Weinberg,
AC CE P
2008), executive functioning (Duschek, Muckenthaler, Werner, & Reyes del Paso, 2009; Hansen, Johnsen, & Thayer, 2003), fluid intelligence (Hinnant, El-Sheikh, Keiley, & Buckhalt, 2013; Staton, El-Sheikh, & Buckhalt, 2009), and cognitive control (Marcovitch et al., 2010; Overbeek, van Boxtel, & Westerink, 2014). Generally, this body of research suggests that a higher level of parasympathetic activity at rest (i.e., higher resting RSA) in conjunction with a decrease in parasympathetic activity in response to a challenge (i.e., RSA withdrawal) is associated with better cognitive performance. For instance, Hinnant et al. (2013) demonstrated that children with lower resting RSA and less RSA withdrawal during tasks measuring fluid cognition showed a significantly shallower slope of fluid cognitive performance across middle childhood. Fairclough et al. (2005) found that greater RSA withdrawal was associated with better
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
6
performance on a demanding learning task and also predicted changes in performance across the task in sample of young adults.
PT
Notably, all of these cognitive tasks with which RSA have been associated are explicit in
RI
nature. Given that executive functions (i.e., attention, working memory) are an integral part of implicit learning and that prefrontal brain regions are active during this type of learning (Pascual-
SC
Leone, Wassermann, Grafman, & Hallett, 1996; Simon, Vaidya, Howard, & Howard, 2012;
NU
Yang & Li, 2012), it stands to reason that RSA may be related to implicit learning in a similar fashion. No research to date has examined the relationship between autonomic activity and
MA
performance on a task that specifically measures implicit learning. This study aims to fill that void by providing a more nuanced understanding of the physiological processes underlying
TE
D
implicit learning, and the possible implications for decision making under uncertainty. We investigated whether RSA measured at rest and during an implicit learning paradigm, the Triplets
AC CE P
Learning Task (TLT; Howard, Howard, Dennis, & Kelly, 2008), were associated with performance on that task. Additionally, we examined the potential interaction between resting RSA and change in RSA during the task (RSA withdrawal) as a predictor of implicit learning. We hypothesized that individuals with higher resting RSA coupled with greater RSA withdrawal during implicit learning would demonstrate the most implicit learning, operationalized as better reaction time-based scores on the TLT. 1. Method 1.1. Participants Seventy-one undergraduate students (37 females, 34 males) from a mid-Atlantic private university participated in this study. Participants were recruited through introductory psychology classes and received course credit as compensation. Data from two male participants were
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
7
discarded because their ages (53 and 60 years old) did not warrant classification as a young adult, resulting in a final sample of 69 participants. This final sample had a mean age of 19.62
PT
years (SD = 2.20), and its racial and ethnic composition closely matched that of the university,
RI
with 55 participants identifying themselves as White (79.7%), three as Black (4.3%), two as Asian (2.9%), eight as Hispanic or Latino (11.6%), and one as Native Hawaiian or Pacific
SC
Islander (1.4%). The university’s Institutional Review Board approved all procedures, and the
NU
participants provided informed consent prior to administration of any measures. 1.2. Procedure
MA
The researcher seated the participant in front of the testing computer and manually adhered nine electrodes to the participant’s skin in order to collect data on autonomic activity.
D
The sensors were connected to a Mindware physiological data collection system, and data on the
TE
participants’ RSA were obtained. The researcher instructed the participant to sit quietly during a
AC CE P
five-minute acclimation period and a subsequent three-minute resting baseline assessment. Then the participant completed a measure of implicit learning, the Triplets Learning Task (TLT; Howard et al., 2008), followed immediately by a two-minute TLT recognition test, described in detail by Howard et al. (2008), to determine if learning was truly implicit. Upon completion, participants were debriefed and received compensation for their time. 1.3. Measures 1.3.1. Triplets Learning Task (TLT). Participants completed the TLT (Howard et al., 2008), a task used to assess implicit probabilistic associative learning in both younger and older adults (e.g., Simon et al., 2012). Individuals viewed a single row of four empty circles on the computer screen and observed events (i.e., a circle filling in red or green) occurring on each trial.
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
8
Each trial was comprised of a three-event sequence, or a “triplet”, consisting of two consecutive red “cues” and a green “target” event, as depicted in Figure 1.
PT
Participants were instructed to observe the first two cue events without responding and to press a key corresponding to the location of the target event. Four buttons on the computer
RI
keyboard were used for responding with the middle and index finger of each hand (“z”, “x”, “.”,
SC
and “/” corresponding to the four circle locations in left-to-right order, respectively). Figure 1
NU
provides a detailed depiction of the timing on each trial of the TLT. Reaction time (RT) was recorded (in milliseconds) from the presentation of the target event to the participant’s initial
MA
response. The target remained on the screen for a fixed amount of time before disappearing. Then an empty row of circles was displayed for 650 ms until the next trial began. Unbeknownst
TE
D
to participants, the location of the first cue predicted one of the four target locations on 80% of the trials (high-probability triplet) and was followed by one of the other three target locations on
AC CE P
20% of the trials (low-probability triplet). The second cue was always randomly selected, thereby creating a second-order probabilistic relationship between the first cue event and the target event. All 64 possible triplets were presented across the trials of the task, with the six unique combinations of cue-target mappings counterbalanced across the sample. Participants were instructed to respond as quickly and accurately as possible to the location of the green circle (target) on each trial. Participants performed two successive sessions, each containing ten blocks of 50 trials for a total of 1,000 trials per person, and so they were encouraged to rest between blocks for no more than 30 seconds to avoid fatigue. Following Simon, Howard, and Howard (2010), end-ofblock feedback was provided to guide participants toward an overall accuracy level of 92%. The triplet type effect (TTE)—the average difference in RT to low-probability and high-probability
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
9
triplets—was calculated for each participant and was used as the indicator of implicit learning in subsequent analyses (Stillman, Feldman, Wambach, Howard, & Howard, 2014). Immediately
PT
after completion of the TLT, explicit awareness was probed using a sensitive recognition test (Howard et al., 2008), during which participants observed 64 triplets in a random order and rated
RI
each as having occurred frequently (“1”) or infrequently (“2”) during the experiment.
SC
1.3.2. Respiratory sinus arrhythmia (RSA). Electrocardiogram (ECG) signals were
NU
recorded from two electrodes, one placed approximately two centimeters (cm) under the participant’s right clavicle and the other between the two bottom ribs on the participant’s left
MA
side, and then analyzed using a commercial software package (MindWare Technologies Ltd., Columbus, OH). Respiration was recorded through four Ag/AgCl ECG electrodes, two of which
TE
D
were placed about three cm apart on the participant’s chest, and two of which were positioned about three cm apart vertically on the mid and lower spine, and was calculated through spectral
AC CE P
analysis of thoracic impedance (Z0; Ernst, Litvack, Lozano, Cacioppo, & Berntson, 1999). Highfrequency HRV, or RSA, was calculated as the changes in the R-R (interbeat) interval occurring between 0.15-0.40 Hz. Minute-by-minute estimates of RSA during baseline and the computerized task were computed through spectral analysis of these heart beat series using a fast Fourier transformation, and correction of errors in the automated marking of R-waves was performed manually by trained coders. The physiological variables used as predictors in subsequent analyses were calculated as the average RSA value during the three-minute baseline period (i.e., resting RSA) and the difference between the average RSA during the TLT and resting RSA (i.e., RSA reactivity). More specifically, a positive reactivity value indicates RSA augmentation during learning while a negative value indicates greater RSA withdrawal during learning.
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
10
2. Results 2.1. Preliminary Analyses
PT
2.1.1. TLT. To facilitate examination of change over time, the TLT was broken down
RI
into four, five-block epochs. Nine RT and accuracy outlier data points (i.e., ≥ 3.5 SD from the mean) were removed, and two types of low-probability triplets—repetitions (e.g., 1-1-1, 2-2-2,
SC
etc.) and trills (e.g., 1-2-1, 2-1-2, etc.)—were removed due to the fact that individuals often show
NU
preexisting response tendencies (Howard et al., 2004). Median RTs were determined by triplet type (high- vs. low-probability) for each participant on each block. The data were then averaged
MA
across adjacent blocks to produce a single mean RT for each triplet type and participant on each of the four epochs. A similar procedure was also used before analyzing the accuracy data.
TE
D
RT data from the TLT was analyzed using a 4 (epoch) x 2 (triplet type) repeated measures analysis of variance (ANOVA), and the data are plotted in Figure 2. Statistical
AC CE P
significance was set at α = .05, and two-tailed tests were used. A Greenhouse-Geisser correction was applied to the degrees of freedom when Mauchly’s test indicated a violation of the assumption of sphericity. This analysis revealed a significant main effect of epoch, F(1.56, 104.63) = 58.56, p < .001, indicating general skill learning or overall faster responding with practice, as well as a significant main effect of triplet type (TT), F(1, 67) = 28.15, p < .001, demonstrating overall sequence-specific learning (i.e., faster responding to high- than lowprobability triplets). The significant epoch x TT interaction illustrated an increased difference in RT to high- and low-probability triplets over the course of the task, F(2.26, 151.20) = 9.57, p < .001. A repeated measures ANOVA on the accuracy data produced a significant main effect of TT, F(1, 64) = 6.02, p = .017, as well as a marginally significant main effect of epoch, F(2.61, 167.02) = 2.37, p = .081, and interaction, F(2.74, 175.03) = 2.62, p = .052. There was no
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
11
evidence of a speed-accuracy trade-off. These findings imply that individuals were more accurate in responding to the high- than low-frequency triplets, and there was a trend for this
PT
difference to increase across the task. Overall, these data suggest that individuals demonstrated
RI
sequence-specific learning on the TLT.
2.1.2. TLT recognition. For the recognition task, reps and trills were included in the
SC
analysis and we examined only responses made to the initial presentation of a particular triplet. A
NU
repeated measures ANOVA produced a marginally significant main effect of TT on frequency ratings, F(3, 207) = 2.27, p = 0.82. Post-hoc paired-samples t-tests revealed a significant
MA
difference between high-probability triplets and repetitions (p = .037) as well as high-probability triplets and trills (p = .020), as expected (Howard et al., 2004). Importantly, the difference
TE
D
between high- and low-probability triplets was not significant, indicating that the learning that occurred during the TLT was implicit in nature.
AC CE P
2.2. Primary Analysis
2.2.1. Descriptive Statistics. Table 1 presents the means and standard deviations for the study variables. As expected, there was a significant negative relationship between resting RSA and RSA reactivity (i.e., RSA during the TLT minus resting RSA), r(63) = -.41, p = .001, suggesting that higher RSA at rest was associated with greater RSA withdrawal during the TLT. TTE scores from the second and third epoch of the TLT were positively related, r(66) = .42, p < .001, while the association between TTE scores from the second and fourth epochs almost reached significance, r(66) = .23, p = .065. There was also a positive relationship between TTE scores on the third and fourth epoch, and r(66) = .39, p = .001. 2.2.2. RSA predicts TLT. We conducted a repeated measures ANOVA to examine whether resting RSA, RSA reactivity and their interaction predicted change over time in learning
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
12
on the TLT. Learning was operationalized as the triplet type effect (TTE)—the average difference between RT to low- and high-probability triplets—and was calculated for each
PT
participant on each of the four epochs. A positive TTE value indicates, on average, faster
RI
responding to high- relative to low-probability triplets. Results showed a significant main effect of RSA reactivity, F(3, 177) = 3.74, p = .012, as well as a resting RSA x RSA reactivity
SC
interaction, F(3, 177) = 3.99, p = .009. Figure 3 displays the patterns of learning over time for
NU
individuals at ± 1 SD from the average values of resting RSA and RSA reactivity. Individuals with higher resting RSA and greater RSA withdrawal demonstrated the greatest learning by the
MA
end of the task. Additionally, the change in TTE across the task for these individuals was steeper than for individuals with other combinations of resting RSA and RSA reactivity.
TE
D
3. Discussion
This study was the first to investigate whether parasympathetic activity at rest and during
AC CE P
an implicit learning task predicted change over time in implicit learning. An understanding of the associations between autonomic activity and implicit learning may have implications for explaining why some individuals are risk-seeking and others are risk-averse under conditions of uncertainty. According to the SMH (Damasio, 1996), learning is the mechanism through which somatic processes have their impact on decision making. Learning that informs decision making likely begins implicitly as the brain subconsciously detects patterns in stimuli and then becomes more explicit as these detected patterns are used to inform higher-level cognitive processes. Through afferent (information to the brain) and efferent (information from the brain) neuronal pathways, in many cases the body reacts to these patterns of stimuli before we are consciously aware of them, creating intuitions or “gut feelings” about situations and behaviors.
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
13
Implicit learning has been shown to play a role in performance on the Iowa Gambling Task (IGT; Bechara, Damasio, Tranel, & Anderson, 1994), in which participants repeatedly draw
PT
from any of four card decks and receive monetary rewards and punishments on each selection
RI
that vary in magnitude and frequency. At the outset of the IGT, participants are unaware that two of the decks will ultimately yield a net gain (i.e., advantageous decks), while the other two will
SC
yield a net loss (i.e., disadvantageous decks). With what begins as trial-and-error, individuals
NU
explore their options and carry forward with them information pertaining to the outcomes they experience. Over the course of the task, learning is illustrated through individuals selecting
MA
significantly more often from the advantageous relative to the disadvantageous decks (Bechara et al., 1994).
TE
D
Previous research with the IGT has shown that individuals exhibit autonomic arousal in anticipation of enacting a decision (Crone, Somsen, van Beek, & van der Molen, 2004; Crone &
AC CE P
van der Molen, 2007; Drucaroff et al., 2011). Bechara, Damasio, Tranel, and Damasio (1997) examined the skin conductance responses (SCRs) of patients with prefrontal damage—a cause of impaired decision making—and healthy controls, and found that only the controls displayed increases in draws from advantageous decks on the IGT. Moreover, only the controls exhibited SCRs before choosing a disadvantageous deck, suggesting that repeated experiences with the choice-outcome contingencies of the various decks enabled healthy participants to form markers in response to them. These markers appeared to guide behavior implicitly, at least initially, as evidenced by the participants’ inability to describe their reasons for preferring certain decks. In support of Bechara et al. (1997), Crone et al. (2004) found that only participants who performed well overall on the IGT (i.e., those who learned) displayed a slowed heart rate and increased SCRs before selecting the disadvantageous decks. While this research provides support for a
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
14
connection between autonomic nervous system activity and learning, this work has primarily focused on SNS activity.
PT
We focused on the role of the PNS in implicit learning due to associations between the
RI
former and more explicit cognitive processes including executive functioning, working memory, and fluid intelligence (Duschek et al., 2009; Fairchild et al., 2005; Haley et al., 2008; Hansen et
SC
al., 2003; Hinnant et al., 2013; Marcovitch et al., 2010; Overbeek et al., 2014; Staton et al.,
NU
2009), as well as the fact that the prefrontal cortex plays a prominent role in PNS activity and in the functional networks active during learning and memory (Thayer et al., 2012; Thayer & Lane,
MA
2000; Yang & Li, 2012). PNS activity was operationalized through RSA, an indirect measure of vagus nerve influence on the heart. High levels of resting RSA and RSA withdrawal are thought
TE
D
to indicate a general physiological readiness and capability to engage with environmental agents (Porges, 2003), largely due to self-regulatory processes initiated by the prefrontal cortex (Thayer
AC CE P
et al., 2012; Thayer & Lane, 2000). Given that the prefrontal cortex also plays a large role in implicit learning (Pascual-Leone et al., 1996; Simon et al., 2012; Yang & Li, 2012), we hypothesized that higher resting RSA and greater RSA withdrawal may promote implicit learning during exposure to subtle, and in this case probabilistic, regularities. We found that young adults with this combination of parasympathetic arousal and reactivity exhibited the steepest slope of learning across the task. While individuals with this pattern of resting RSA and RSA reactivity in the current study demonstrated approximately equivalent implicit learning to others early on, as they continued to subconsciously observe the regularities in the task, they exhibited stronger performance. By the third epoch of task, these individuals evidenced greater learning than those with other combinations of resting RSA and RSA reactivity. Importantly, the recognition task confirmed that learning on the task was implicit
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
15
as evidenced by the failure of participants to differentiate high- and low-probability triplets. Overall, this suggests that RSA is not only an index of physiological readiness for the
PT
environment, but psychological as well.
RI
This is the first study to illustrate that any index of autonomic activity is specifically associated with performance on an implicit learning task. Repeated decisions made under
SC
uncertainty (i.e., with probabilistic outcomes) necessarily involve implicit learning as individuals
NU
begin to discern patterns in the data and to make use of those patterns to inform future decisions. The Iowa Gambling Task is a prime example of such a condition in research, though implicit
MA
learning, explicit learning, and decision making are difficult to differentiate in this measure. Mata, Josef, Samanez-Larkin, and Hertwig (2011) explain that learning choice-outcome
TE
D
contingencies over time is a crucial part of decision making under uncertainty. This learning may occur implicitly so that people can acquire information about the environment and make
AC CE P
advantageous decisions without exerting conscious effort (Frensch & Rünger, 2003). Our findings suggest that PNS activity, as measured by RSA, may play an important role as a somatic marker in implicit learning, and that patterns of higher resting PNS levels and greater PNS reactivity may facilitate implicit learning processes. These same processes may play a key role in decision making under conditions of uncertainty by making some individuals more sensitive to or better able to detect patterns in stimuli, and therefore better able to adapt or regulate their behavior. While this study addresses a major gap in the existing literature by examining the relations between autonomic activity and implicit learning, it does have some limitations. One limitation is our inability to document autonomic reactivity to high- and low-probability triplets separately. The temporal resolution of autonomic measures is incompatible with the fluid, event-
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
16
based design of the implicit learning measure, the TLT. Thus, innovative future work might investigate the feasibility of discerning the effects of autonomic indices on learning under
PT
different conditions of stimuli frequency (i.e., do autonomic activity effects differ depending on
RI
the familiarity of the event). Future research should also consider developmental aspects of implicit learning and how links between autonomic processes and implicit learning change over
SC
time in order to better understand how decision-making processes develop. Previous research has
NU
demonstrated age differences in implicit learning across adulthood (Mata et al., 2011). However, little is currently known about changes in implicit learning across childhood and adolescence,
MA
except for conclusions inferred from work using tasks that are not entirely implicit, such as the IGT (e.g., Crone et al., 2004). While resting RSA appears to be moderately stable in childhood,
TE
D
RSA reactivity shows less stability (El-Sheikh, 2005). Longitudinal work investigating RSA stability during adolescence to early adulthood and its association with implicit learning could
AC CE P
help to clarify whether RSA plays a role in the dramatic increases in risky decision making and risk-taking behaviors that occur during this often emotionally stressful period.
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
17
References Bechara, A., & Damasio, A.R. (2005). The somatic marker hypothesis: A neural theory of
PT
economic decision. Games and Economic Behavior, 52(2), 336-372.
RI
doi:10.1016/j.geb.2004.06.010
Bechara, A., Damasio, A.R., Tranel, D., & Anderson, S. (1994). Insensitivity to future
SC
consequences following damage to human prefrontal cortex. Cognition, 50(1), 7-15.
NU
doi:10.1016/0010-0277(94)90018-3
Berntson, G.G., Bigger, J.T., Jr., Eckberg, D.L., Grossman, P., Kaufmann, P.G., Malik, M.,
MA
Nagaraja, H.N.,…van der Molen, M.W. (1997). Heart rate variability: Origins, methods, and interpretive caveats. Psychophysiology, 34(6), 623-648.
TE
D
doi:10.1111/j.14698986.1997.tb02140.x Crone, E.A., Somsen, R.J.M., van Beek, B., & van der Molen, M.W. (2004). Heart rate and skin
AC CE P
conductance analysis of antecedents and consequences of decision making. Psychophysiology, 41, 531-540. Crone, E.A., & van der Molen, M.W. (2007). Development of decision making in school-aged children and adolescents: Evidence from heart rate and skin conductance analysis. Child Development, 78(4), 1288-1301. doi:10.1111/j.1467-8624.2007.0106 Damasio, A.R. (1996). The somatic marker hypothesis and the possible functions of the prefrontal cortex. Philosophical Transactions, 351(1346), 1413-1420. doi:10.1098/rstb.1996.0125 Drucaroff, L.J., Kievit, R., Guinjoan, S.M., Gerschcovich, E.R., Cerquetti, D., Leiguarda, R.,…Vigo, D.E. (2011). Higher autonomic activation predicts better performance in Iowa gambling task. Cognitive and Behavioral Neurology, 24(2), 93-98.
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
18
doi:10.1097/WNN.0b013e3182239308 Dunn, B.D., Dalgleish, T., & Lawrence, A.D. (2006). The somatic marker hypothesis: A critical
PT
evaluation. Neuroscience and Biobehavioral Reviews, 30(2), 239-271.
RI
doi:10.1016/j.neubiorev.2005.07.001
Duschek, S., Muckenthaler, M., Werner, N., & Reyes del Paso, G.A. (2009). Relationships
SC
between features of autonomic cardiovascular control and cognitive performance.
NU
Biological Psychology, 81, 110-117. doi:10.1016/j.biopsycho.2009.03.003 El-Sheikh, M. (2005). Stability of respiratory sinus arrhythmia in children and young
MA
adolescents: A longitudinal examination. Developmental Psychobiology, 46(1), 66-74. doi:10.1002/dev.20036
TE
D
Ernst, J.M., Litvack, D.A., Lozano, D.L., Cacioppo, J.T., & Berntson, G.G. (1999). Impedance pneumography: Noise as signal in impedance cardiography. Psychophysiology, 36(3),
AC CE P
333-338. doi:10.1017/S0048577299981003 Fairclough, S.H., Venables, L., & Tattersall, A. (2005). The influence of task demand and learning on the psychophysiological response. International Journal of Psychophysiology, 56, 171-184. doi:10.1016/j.ijpsycho.2004.11.003 Frazier, T.W., Strauss, M.E., & Steinhauer, S.R. (2004). Respiratory sinus arrhythmia as an index of emotional response in young adults. Psychophysiology, 41(1), 75-83. doi:10.1046/j.1469-8986.2003.0013 Frensch, P.A., & Rünger, D. (2003). Implicit Learning. Current Directions in Psychological Science, 12(1), 13-18. doi:10.1111/1467-8721.01213
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
19
Haley, D.W., Grunau, R.E., Oberlander, T.F., & Weinberg, J. (2008). Contingency learning and reactivity in preterm and full-term infants at 3 months. Infancy, 13(6), 570-595.
PT
doi:10.1080/15250000802458682
RI
Hansen, A.L., Johnson, B.H., & Thayer, J.F. (2003). Vagal influence on working memory and attention. International Journal of Psychophysiology, 48, 263-274. doi:10.1016/S0167-
SC
8760(03)00073-4
NU
Hinnant, J.B., El-Sheikh, M., Keiley, M., & Buckhalt, J.A. (2013). Marital conflict, allostatic load, and the development of children’s fluid cognitive performance. Child Development,
MA
84(6), 2003-2014. doi:10.1111/cdev.12103
Howard, J.H., Jr., Howard, D.V., Dennis, N.A., & Kelly, A.J. (2008). Implicit learning of
TE
D
predictive relationships in three-element visual sequences by young and old adults. Journal of Experimental Psychology: Learning, Memory & Cognition, 34(5), 1139-1157.
AC CE P
doi:10.1037/a0012797
Marcovitch, S., Leigh, J., Calkins, S.D., Leerks, E.M., O’Brien, M., & Blankson, A.M. (2010). Moderate vagal withdrawal in 3.5-year-old children is associated with optimal performance on executive function tasks. Developmental Psychobiology, 52(6), 603-608. doi:10.1002/dev.20462
Mata, R., Josef, A.K., Samanez-Larkin, G.R., & Hertwig, R. (2011). Age differences in risky choice: A meta-analysis. Annals of the NY Academy of Sciences, 1235(1), 18-29. doi:10.1111/j.1749-6632.2011.0620 Overbeek, T.J.M., van Boxtel, T., & Westerink J.H.D.M. (2014). Respiratory sinus arrhythmia responses to cognitive tasks: Effects of task factors and RSA indices. Biological Psychology, 99, 1-14. doi:10.1016/j.biopsycho.2014.02.006
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
20
Pascual-Leone, A., Wassermann, E.M., Grafman, J., & Hallett, M. (1996). The role of the dorsolateral prefrontal cortex in implicit procedural learning. Experimental Brain
PT
Research, 107(3), 479-485.
RI
Porges, S.W. (1992). Vagal tone: A physiological marker of stress vulnerability. Pediatrics, 90(3), 498-504.
SC
Porges, S.W. (2003). The polyvagal theory: phylogenetic contributions to social behavior.
NU
Physiology & Behavior, 79(3), 503-513. doi:10.1016/S0167-8760(01)00162-3 Porges, S.W. (2007). The polyvagal perspective. Biological Psychology, 74(2), 116-143.
MA
doi:10.1016/j.biopsycho.2006.06.009
Reber, A.S. (1989). Implicit learning and tacit knowledge. Journal of Experimental Psychology:
TE
D
General, 118(3), 219-235. doi:10.1037/0096-3445.118.3.219 Simon, J.R., Howard, J.H., Jr., & Howard, D.V. (2010). Adult age differences in learning from
AC CE P
positive and negative probabilistic feedback. Neuropsychology, 24(4), 534-541. doi:10.1037/a0018652
Simon, J.R., Vaidya, C.J., Howard, J.H., Jr., & Howard, D.V. (2012). The effects of aging on the neural basis of implicit associative learning in a probabilistic triplets learning task. Journal of Cognitive Neuroscience, 24(2), 451-463. doi:10.1162/jocn_a_00116 Staton, L., El-Sheikh, M., & Buckhalt, J.A. (2009). Respiratory sinus arrhythmia and cognitive functioning in children. Developmental Psychobiology, 51(3), 249-258. doi:10.1002/dev.20361 Stillman, C.M., Feldman, H., Wambach, C.G., Howard, J.H., Jr., & Howard, D.V. (2014). Dispositional mindfulness is associated with reduced implicit learning. Consciousness and Cognition, 28, 141-150. doi:10.1016/j.concog.2014.07.002
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
21
Thayer, J.F., Ahs, F., Fredrikson, M., Sollers, J.J., III. & Wager, T.D. (2012). A meta-analysis of heart rate variability and neuroimaging studies: Implications for heart rate variability as a
PT
marker of stress and health. Neuroscience and Biobehavioral Reviews, 36, 747-756.
RI
doi:10.1016/j.neubiorev.2011.11.009
Thayer, J.F., & Lane, R.D. (2000). A model of neurovisceral integration in emotion regulation
SC
and dysregulation. Journal of Affective Disorders, 61, 201-216.
NU
Yang, J., & Li, P. (2012). Brain networks of explicit and implicit learning. PLoS One, 7(8),
AC CE P
TE
D
MA
e42993. doi:10.1371/journal.pone.0042993
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
22
Cue 1
PT
Cue 2
RI
Target
SC
270 ms
NU
270 ms
MA
Respond
D
Figure 1. An example of a trial on the TLT. Participants observe the first two events and
AC CE P
TE
respond only to the location of the target event.
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING 460
440
PT
430 420 410
RI
High Probability Low Probability
SC
400 390 380 370 360 2
3 Epoch
MA
1
NU
Mean Median RT (in msec)
450
4
D
Figure 2. Mean of median RT (in milliseconds) over epochs by high- and low-probability
AC CE P
TE
triplets. Error bars represent the standard error of the mean.
23
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
25
PT
High Resting RSA; Less RSA withdrawal
15
High Resting RSA; Greater RSA withdrawal
5 0 Epoch 1
Epoch 2
Epoch 3
Epoch 4
SC
RI
10
NU
Triplet Type Effect (in ms)
20
Low Resting RSA; Greater RSA withdrawal
MA
-5
Low Resting RSA; Less RSA withdrawal
-10
TE
D
Figure 3. The interaction of resting RSA and RSA reactivity on learning across the TLT. A
AC CE P
larger triplet type effect (TTE) indicates greater learning.
24
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
RI
PT
Table 1. Descriptive statistics for study variables. Variable M SD Resting RSA 6.10 1.05 RSA Reactivity -0.06 0.60 TTE Epoch 1 -0.89 28.29 TTE Epoch 2 10.92 18.82 TTE Epoch 3 8.75 18.34 TTE Epoch 4 16.37 20.99
25
SC
Note. RSA Reactivity was calculated as the difference between average RSA during the
NU
Triplets Learning Task (TLT) and resting RSA. Therefore, a positive reactivity value denotes an increase in RSA from baseline to the task. TTE was calculated as the average difference in
MA
reaction time to low- and high-probability triplets. Thus, a positive TTE value indicates, on average, faster responding to high- relative to low-probability triplets. RSA = respiratory sinus
AC CE P
TE
D
arrhythmia. TTE = triplet type effect.
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING Highlights
TE
D
MA
NU
SC
RI
PT
Respiratory sinus arrhythmia (RSA) predicts change over time in implicit learning Greater resting RSA and RSA withdrawal predicts better implicit learning Links between RSA and learning may help explain decision making under uncertainty
AC CE P
26
S0167-8760(16)30738-3 doi: 10.1016/j.ijpsycho.2016.10.014 INTPSY 11191
To appear in:
International Journal of Psychophysiology
Received date: Revised date: Accepted date:
30 June 2016 10 October 2016 17 October 2016
Please cite this article as: Forman-Alberti, Alissa B., Hinnant, J. Benjamin, Links between autonomic activity and implicit learning, International Journal of Psychophysiology (2016), doi: 10.1016/j.ijpsycho.2016.10.014
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
NU
SC
RI
PT
Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
MA
Links between Autonomic Activity and Implicit Learning
Alissa B. Forman-Albertia and J. Benjamin Hinnantb Catholic University of America
D
a
AC CE P
TE
b
Auburn University
Author Note
Alissa B. Forman-Alberti, Department of Psychology, The Catholic University of America; J. Benjamin Hinnant, Department of Human Development and Family Studies, Auburn University Address correspondence to Alissa B. Forman-Alberti, Department of Psychology, 255 O’Boyle Hall, The Catholic University of America, 620 Michigan Ave NE, Washington, DC 20064. Email: [email protected]
1
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
2
Abstract The somatic marker hypothesis posits that autonomic activity occurring in response to specific
PT
stimuli aids in implicit learning, the learning of information without explicit awareness of what
RI
has been learned. This study investigated whether respiratory sinus arrhythmia (RSA), a measure of autonomic nervous system activity, predicted changes in implicit learning. The interaction of
SC
resting RSA and RSA reactivity (change in RSA during the implicit learning task) was
NU
associated with changes in implicit learning, with those who had higher resting RSA and greater RSA withdrawal during the task performing better. These findings contribute to a better
MA
understanding of the autonomic processes that may underlie implicit learning and are discussed in relation to potential links between autonomic activity, implicit learning, and decision making.
AC CE P
making
TE
D
Keywords: autonomic activity; respiratory sinus arrhythmia; implicit learning; decision
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
3
Links between Autonomic Activity and Implicit Learning We learn even when we are not aware of learning taking place. Implicit learning is the
PT
process of gaining sensitivity to novel relationships or regularities among events in the
RI
environment, which are often probabilistic in nature, without conscious knowledge of what has been learned or that learning has even occurred (Reber, 1989). Implicit learning permits
SC
individuals to adapt their behavior to environments without expending conscious effort (Frensch
NU
& Rünger, 2003) and may be experienced as intuitions or “gut feelings” about the available choices (Lo & Repin, 2002). In other words, even though we may not be consciously aware of
MA
learning, our bodies may be subtly influencing our responses to patterns in environmental stimuli.
TE
D
According to the somatic marker hypothesis (SMH; Bechara & Damasio, 2005; Damasio, 1996), autonomic activity plays a key role in implicit learning processes. The SMH suggests that
AC CE P
physiological “markers” occur within individuals in response to encounters with stimuli and that with repeated exposure, these markers then become associated with those events. Somatic markers will be reactivated in response to those stimuli in future experiences, even without conscious awareness of the stimuli’s significance, and will bias cognitive processing and behavior involving them (for a review, see Dunn, Dalgleish, & Lawrence, 2006). Existing research supports the claim that somatic signals, at least initially, assist with learning in uncertain environments, but the majority of this research has focused only on activity in the sympathetic branch of the autonomic nervous system, and has largely ignored the parasympathetic branch. The relationship between the brain and the autonomic nervous system is crucial to understanding the connection between the latter and individuals’ behavior. Both cortical and subcortical areas of the brain contribute to processes such as attentional regulation, affective
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
4
information processing, and behavioral flexibility (Thayer & Lane, 2000). The coordinated activation of all of these processes is what allows individuals to adapt their behavior in the face
PT
of environmental challenges or changes (Thayer, Ahs, Fredrikson, Sollers, & Wager, 2012). For
RI
example, implicit learning involves a functional network of cortical-subcortical connections, with the prefrontal cortex acting as a hub that links brain regions related to memory and learning,
SC
such as the caudate nucleus (Yang & Li, 2012). In order to adapt behavior properly, there needs
NU
to be an appropriate balance between the sympathetic and parasympathetic nervous system (SNS and PNS, respectively); the former stimulates the body during stressful situations when a fight-
MA
or-flight response is needed and the latter promotes readiness to engage with (e.g., learn from) the environment. Research has suggested that this ability to regulate autonomic activity may be
TE
D
related to psychological self-regulation, which also relies on the prefrontal cortex to exert topdown control on other brain regions (Porges, 1992). Given this, PNS activity may be particularly
functioning.
AC CE P
important to consider when examining individual differences in higher-order cognitive
The current study focused on PNS activity and its role as a potential somatic marker in implicit learning. Although this is the first study on this topic, there is evidence that the PNS plays an important role in social and more explicit learning processes (Porges, 2007). To measure PNS activity, the impact of the vagus nerve, one of the primary conduits for parasympathetic influence on multiple organ systems, can be non-invasively assessed. Output from the brain, the nature of which is dependent on an individual’s internal and external circumstances, travels to the heart via the vagus nerve. The vagus nerve acts like a “brake” on the heart (Porges, 2007), increasing cardiac output to facilitate blood flow for active problem solving or promoting fight-or-flight response (i.e., SNS activation) when the brake is withdrawn.
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
5
The ability to flexibly engage and withdraw parasympathetic influence depending on internal and external demands is indicative of adaptive attentional and regulatory processes (Porges, 2003).
PT
This interplay of sympathetic and parasympathetic influence on the heart leads to variability in
RI
heart rate (i.e., heart rate variability or HRV). Because vagal activity fluctuates in phase with respiration (Frazier, Strauss, & Steinhauer, 2004), researchers can document the variability in
SC
inter-beat intervals (i.e., the time between heartbeats) that occurs during typical breathing cycles,
NU
also known as high-frequency HRV (Berntson et al., 1997). This high-frequency HRV is also known as respiratory sinus arrhythmia (RSA). Therefore, RSA acts as an indirect measure of
MA
vagal-parasympathetic influence on the heart.
There is extensive research linking both RSA at rest as well as changes in RSA in
TE
D
response to a stressor (i.e., RSA reactivity) to a host of cognitive outcomes including explicit learning (Fairclough, Venables, & Tattersall, 2005; Haley, Grunau, Oberlander, & Weinberg,
AC CE P
2008), executive functioning (Duschek, Muckenthaler, Werner, & Reyes del Paso, 2009; Hansen, Johnsen, & Thayer, 2003), fluid intelligence (Hinnant, El-Sheikh, Keiley, & Buckhalt, 2013; Staton, El-Sheikh, & Buckhalt, 2009), and cognitive control (Marcovitch et al., 2010; Overbeek, van Boxtel, & Westerink, 2014). Generally, this body of research suggests that a higher level of parasympathetic activity at rest (i.e., higher resting RSA) in conjunction with a decrease in parasympathetic activity in response to a challenge (i.e., RSA withdrawal) is associated with better cognitive performance. For instance, Hinnant et al. (2013) demonstrated that children with lower resting RSA and less RSA withdrawal during tasks measuring fluid cognition showed a significantly shallower slope of fluid cognitive performance across middle childhood. Fairclough et al. (2005) found that greater RSA withdrawal was associated with better
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
6
performance on a demanding learning task and also predicted changes in performance across the task in sample of young adults.
PT
Notably, all of these cognitive tasks with which RSA have been associated are explicit in
RI
nature. Given that executive functions (i.e., attention, working memory) are an integral part of implicit learning and that prefrontal brain regions are active during this type of learning (Pascual-
SC
Leone, Wassermann, Grafman, & Hallett, 1996; Simon, Vaidya, Howard, & Howard, 2012;
NU
Yang & Li, 2012), it stands to reason that RSA may be related to implicit learning in a similar fashion. No research to date has examined the relationship between autonomic activity and
MA
performance on a task that specifically measures implicit learning. This study aims to fill that void by providing a more nuanced understanding of the physiological processes underlying
TE
D
implicit learning, and the possible implications for decision making under uncertainty. We investigated whether RSA measured at rest and during an implicit learning paradigm, the Triplets
AC CE P
Learning Task (TLT; Howard, Howard, Dennis, & Kelly, 2008), were associated with performance on that task. Additionally, we examined the potential interaction between resting RSA and change in RSA during the task (RSA withdrawal) as a predictor of implicit learning. We hypothesized that individuals with higher resting RSA coupled with greater RSA withdrawal during implicit learning would demonstrate the most implicit learning, operationalized as better reaction time-based scores on the TLT. 1. Method 1.1. Participants Seventy-one undergraduate students (37 females, 34 males) from a mid-Atlantic private university participated in this study. Participants were recruited through introductory psychology classes and received course credit as compensation. Data from two male participants were
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
7
discarded because their ages (53 and 60 years old) did not warrant classification as a young adult, resulting in a final sample of 69 participants. This final sample had a mean age of 19.62
PT
years (SD = 2.20), and its racial and ethnic composition closely matched that of the university,
RI
with 55 participants identifying themselves as White (79.7%), three as Black (4.3%), two as Asian (2.9%), eight as Hispanic or Latino (11.6%), and one as Native Hawaiian or Pacific
SC
Islander (1.4%). The university’s Institutional Review Board approved all procedures, and the
NU
participants provided informed consent prior to administration of any measures. 1.2. Procedure
MA
The researcher seated the participant in front of the testing computer and manually adhered nine electrodes to the participant’s skin in order to collect data on autonomic activity.
D
The sensors were connected to a Mindware physiological data collection system, and data on the
TE
participants’ RSA were obtained. The researcher instructed the participant to sit quietly during a
AC CE P
five-minute acclimation period and a subsequent three-minute resting baseline assessment. Then the participant completed a measure of implicit learning, the Triplets Learning Task (TLT; Howard et al., 2008), followed immediately by a two-minute TLT recognition test, described in detail by Howard et al. (2008), to determine if learning was truly implicit. Upon completion, participants were debriefed and received compensation for their time. 1.3. Measures 1.3.1. Triplets Learning Task (TLT). Participants completed the TLT (Howard et al., 2008), a task used to assess implicit probabilistic associative learning in both younger and older adults (e.g., Simon et al., 2012). Individuals viewed a single row of four empty circles on the computer screen and observed events (i.e., a circle filling in red or green) occurring on each trial.
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
8
Each trial was comprised of a three-event sequence, or a “triplet”, consisting of two consecutive red “cues” and a green “target” event, as depicted in Figure 1.
PT
Participants were instructed to observe the first two cue events without responding and to press a key corresponding to the location of the target event. Four buttons on the computer
RI
keyboard were used for responding with the middle and index finger of each hand (“z”, “x”, “.”,
SC
and “/” corresponding to the four circle locations in left-to-right order, respectively). Figure 1
NU
provides a detailed depiction of the timing on each trial of the TLT. Reaction time (RT) was recorded (in milliseconds) from the presentation of the target event to the participant’s initial
MA
response. The target remained on the screen for a fixed amount of time before disappearing. Then an empty row of circles was displayed for 650 ms until the next trial began. Unbeknownst
TE
D
to participants, the location of the first cue predicted one of the four target locations on 80% of the trials (high-probability triplet) and was followed by one of the other three target locations on
AC CE P
20% of the trials (low-probability triplet). The second cue was always randomly selected, thereby creating a second-order probabilistic relationship between the first cue event and the target event. All 64 possible triplets were presented across the trials of the task, with the six unique combinations of cue-target mappings counterbalanced across the sample. Participants were instructed to respond as quickly and accurately as possible to the location of the green circle (target) on each trial. Participants performed two successive sessions, each containing ten blocks of 50 trials for a total of 1,000 trials per person, and so they were encouraged to rest between blocks for no more than 30 seconds to avoid fatigue. Following Simon, Howard, and Howard (2010), end-ofblock feedback was provided to guide participants toward an overall accuracy level of 92%. The triplet type effect (TTE)—the average difference in RT to low-probability and high-probability
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
9
triplets—was calculated for each participant and was used as the indicator of implicit learning in subsequent analyses (Stillman, Feldman, Wambach, Howard, & Howard, 2014). Immediately
PT
after completion of the TLT, explicit awareness was probed using a sensitive recognition test (Howard et al., 2008), during which participants observed 64 triplets in a random order and rated
RI
each as having occurred frequently (“1”) or infrequently (“2”) during the experiment.
SC
1.3.2. Respiratory sinus arrhythmia (RSA). Electrocardiogram (ECG) signals were
NU
recorded from two electrodes, one placed approximately two centimeters (cm) under the participant’s right clavicle and the other between the two bottom ribs on the participant’s left
MA
side, and then analyzed using a commercial software package (MindWare Technologies Ltd., Columbus, OH). Respiration was recorded through four Ag/AgCl ECG electrodes, two of which
TE
D
were placed about three cm apart on the participant’s chest, and two of which were positioned about three cm apart vertically on the mid and lower spine, and was calculated through spectral
AC CE P
analysis of thoracic impedance (Z0; Ernst, Litvack, Lozano, Cacioppo, & Berntson, 1999). Highfrequency HRV, or RSA, was calculated as the changes in the R-R (interbeat) interval occurring between 0.15-0.40 Hz. Minute-by-minute estimates of RSA during baseline and the computerized task were computed through spectral analysis of these heart beat series using a fast Fourier transformation, and correction of errors in the automated marking of R-waves was performed manually by trained coders. The physiological variables used as predictors in subsequent analyses were calculated as the average RSA value during the three-minute baseline period (i.e., resting RSA) and the difference between the average RSA during the TLT and resting RSA (i.e., RSA reactivity). More specifically, a positive reactivity value indicates RSA augmentation during learning while a negative value indicates greater RSA withdrawal during learning.
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
10
2. Results 2.1. Preliminary Analyses
PT
2.1.1. TLT. To facilitate examination of change over time, the TLT was broken down
RI
into four, five-block epochs. Nine RT and accuracy outlier data points (i.e., ≥ 3.5 SD from the mean) were removed, and two types of low-probability triplets—repetitions (e.g., 1-1-1, 2-2-2,
SC
etc.) and trills (e.g., 1-2-1, 2-1-2, etc.)—were removed due to the fact that individuals often show
NU
preexisting response tendencies (Howard et al., 2004). Median RTs were determined by triplet type (high- vs. low-probability) for each participant on each block. The data were then averaged
MA
across adjacent blocks to produce a single mean RT for each triplet type and participant on each of the four epochs. A similar procedure was also used before analyzing the accuracy data.
TE
D
RT data from the TLT was analyzed using a 4 (epoch) x 2 (triplet type) repeated measures analysis of variance (ANOVA), and the data are plotted in Figure 2. Statistical
AC CE P
significance was set at α = .05, and two-tailed tests were used. A Greenhouse-Geisser correction was applied to the degrees of freedom when Mauchly’s test indicated a violation of the assumption of sphericity. This analysis revealed a significant main effect of epoch, F(1.56, 104.63) = 58.56, p < .001, indicating general skill learning or overall faster responding with practice, as well as a significant main effect of triplet type (TT), F(1, 67) = 28.15, p < .001, demonstrating overall sequence-specific learning (i.e., faster responding to high- than lowprobability triplets). The significant epoch x TT interaction illustrated an increased difference in RT to high- and low-probability triplets over the course of the task, F(2.26, 151.20) = 9.57, p < .001. A repeated measures ANOVA on the accuracy data produced a significant main effect of TT, F(1, 64) = 6.02, p = .017, as well as a marginally significant main effect of epoch, F(2.61, 167.02) = 2.37, p = .081, and interaction, F(2.74, 175.03) = 2.62, p = .052. There was no
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
11
evidence of a speed-accuracy trade-off. These findings imply that individuals were more accurate in responding to the high- than low-frequency triplets, and there was a trend for this
PT
difference to increase across the task. Overall, these data suggest that individuals demonstrated
RI
sequence-specific learning on the TLT.
2.1.2. TLT recognition. For the recognition task, reps and trills were included in the
SC
analysis and we examined only responses made to the initial presentation of a particular triplet. A
NU
repeated measures ANOVA produced a marginally significant main effect of TT on frequency ratings, F(3, 207) = 2.27, p = 0.82. Post-hoc paired-samples t-tests revealed a significant
MA
difference between high-probability triplets and repetitions (p = .037) as well as high-probability triplets and trills (p = .020), as expected (Howard et al., 2004). Importantly, the difference
TE
D
between high- and low-probability triplets was not significant, indicating that the learning that occurred during the TLT was implicit in nature.
AC CE P
2.2. Primary Analysis
2.2.1. Descriptive Statistics. Table 1 presents the means and standard deviations for the study variables. As expected, there was a significant negative relationship between resting RSA and RSA reactivity (i.e., RSA during the TLT minus resting RSA), r(63) = -.41, p = .001, suggesting that higher RSA at rest was associated with greater RSA withdrawal during the TLT. TTE scores from the second and third epoch of the TLT were positively related, r(66) = .42, p < .001, while the association between TTE scores from the second and fourth epochs almost reached significance, r(66) = .23, p = .065. There was also a positive relationship between TTE scores on the third and fourth epoch, and r(66) = .39, p = .001. 2.2.2. RSA predicts TLT. We conducted a repeated measures ANOVA to examine whether resting RSA, RSA reactivity and their interaction predicted change over time in learning
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
12
on the TLT. Learning was operationalized as the triplet type effect (TTE)—the average difference between RT to low- and high-probability triplets—and was calculated for each
PT
participant on each of the four epochs. A positive TTE value indicates, on average, faster
RI
responding to high- relative to low-probability triplets. Results showed a significant main effect of RSA reactivity, F(3, 177) = 3.74, p = .012, as well as a resting RSA x RSA reactivity
SC
interaction, F(3, 177) = 3.99, p = .009. Figure 3 displays the patterns of learning over time for
NU
individuals at ± 1 SD from the average values of resting RSA and RSA reactivity. Individuals with higher resting RSA and greater RSA withdrawal demonstrated the greatest learning by the
MA
end of the task. Additionally, the change in TTE across the task for these individuals was steeper than for individuals with other combinations of resting RSA and RSA reactivity.
TE
D
3. Discussion
This study was the first to investigate whether parasympathetic activity at rest and during
AC CE P
an implicit learning task predicted change over time in implicit learning. An understanding of the associations between autonomic activity and implicit learning may have implications for explaining why some individuals are risk-seeking and others are risk-averse under conditions of uncertainty. According to the SMH (Damasio, 1996), learning is the mechanism through which somatic processes have their impact on decision making. Learning that informs decision making likely begins implicitly as the brain subconsciously detects patterns in stimuli and then becomes more explicit as these detected patterns are used to inform higher-level cognitive processes. Through afferent (information to the brain) and efferent (information from the brain) neuronal pathways, in many cases the body reacts to these patterns of stimuli before we are consciously aware of them, creating intuitions or “gut feelings” about situations and behaviors.
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
13
Implicit learning has been shown to play a role in performance on the Iowa Gambling Task (IGT; Bechara, Damasio, Tranel, & Anderson, 1994), in which participants repeatedly draw
PT
from any of four card decks and receive monetary rewards and punishments on each selection
RI
that vary in magnitude and frequency. At the outset of the IGT, participants are unaware that two of the decks will ultimately yield a net gain (i.e., advantageous decks), while the other two will
SC
yield a net loss (i.e., disadvantageous decks). With what begins as trial-and-error, individuals
NU
explore their options and carry forward with them information pertaining to the outcomes they experience. Over the course of the task, learning is illustrated through individuals selecting
MA
significantly more often from the advantageous relative to the disadvantageous decks (Bechara et al., 1994).
TE
D
Previous research with the IGT has shown that individuals exhibit autonomic arousal in anticipation of enacting a decision (Crone, Somsen, van Beek, & van der Molen, 2004; Crone &
AC CE P
van der Molen, 2007; Drucaroff et al., 2011). Bechara, Damasio, Tranel, and Damasio (1997) examined the skin conductance responses (SCRs) of patients with prefrontal damage—a cause of impaired decision making—and healthy controls, and found that only the controls displayed increases in draws from advantageous decks on the IGT. Moreover, only the controls exhibited SCRs before choosing a disadvantageous deck, suggesting that repeated experiences with the choice-outcome contingencies of the various decks enabled healthy participants to form markers in response to them. These markers appeared to guide behavior implicitly, at least initially, as evidenced by the participants’ inability to describe their reasons for preferring certain decks. In support of Bechara et al. (1997), Crone et al. (2004) found that only participants who performed well overall on the IGT (i.e., those who learned) displayed a slowed heart rate and increased SCRs before selecting the disadvantageous decks. While this research provides support for a
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
14
connection between autonomic nervous system activity and learning, this work has primarily focused on SNS activity.
PT
We focused on the role of the PNS in implicit learning due to associations between the
RI
former and more explicit cognitive processes including executive functioning, working memory, and fluid intelligence (Duschek et al., 2009; Fairchild et al., 2005; Haley et al., 2008; Hansen et
SC
al., 2003; Hinnant et al., 2013; Marcovitch et al., 2010; Overbeek et al., 2014; Staton et al.,
NU
2009), as well as the fact that the prefrontal cortex plays a prominent role in PNS activity and in the functional networks active during learning and memory (Thayer et al., 2012; Thayer & Lane,
MA
2000; Yang & Li, 2012). PNS activity was operationalized through RSA, an indirect measure of vagus nerve influence on the heart. High levels of resting RSA and RSA withdrawal are thought
TE
D
to indicate a general physiological readiness and capability to engage with environmental agents (Porges, 2003), largely due to self-regulatory processes initiated by the prefrontal cortex (Thayer
AC CE P
et al., 2012; Thayer & Lane, 2000). Given that the prefrontal cortex also plays a large role in implicit learning (Pascual-Leone et al., 1996; Simon et al., 2012; Yang & Li, 2012), we hypothesized that higher resting RSA and greater RSA withdrawal may promote implicit learning during exposure to subtle, and in this case probabilistic, regularities. We found that young adults with this combination of parasympathetic arousal and reactivity exhibited the steepest slope of learning across the task. While individuals with this pattern of resting RSA and RSA reactivity in the current study demonstrated approximately equivalent implicit learning to others early on, as they continued to subconsciously observe the regularities in the task, they exhibited stronger performance. By the third epoch of task, these individuals evidenced greater learning than those with other combinations of resting RSA and RSA reactivity. Importantly, the recognition task confirmed that learning on the task was implicit
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
15
as evidenced by the failure of participants to differentiate high- and low-probability triplets. Overall, this suggests that RSA is not only an index of physiological readiness for the
PT
environment, but psychological as well.
RI
This is the first study to illustrate that any index of autonomic activity is specifically associated with performance on an implicit learning task. Repeated decisions made under
SC
uncertainty (i.e., with probabilistic outcomes) necessarily involve implicit learning as individuals
NU
begin to discern patterns in the data and to make use of those patterns to inform future decisions. The Iowa Gambling Task is a prime example of such a condition in research, though implicit
MA
learning, explicit learning, and decision making are difficult to differentiate in this measure. Mata, Josef, Samanez-Larkin, and Hertwig (2011) explain that learning choice-outcome
TE
D
contingencies over time is a crucial part of decision making under uncertainty. This learning may occur implicitly so that people can acquire information about the environment and make
AC CE P
advantageous decisions without exerting conscious effort (Frensch & Rünger, 2003). Our findings suggest that PNS activity, as measured by RSA, may play an important role as a somatic marker in implicit learning, and that patterns of higher resting PNS levels and greater PNS reactivity may facilitate implicit learning processes. These same processes may play a key role in decision making under conditions of uncertainty by making some individuals more sensitive to or better able to detect patterns in stimuli, and therefore better able to adapt or regulate their behavior. While this study addresses a major gap in the existing literature by examining the relations between autonomic activity and implicit learning, it does have some limitations. One limitation is our inability to document autonomic reactivity to high- and low-probability triplets separately. The temporal resolution of autonomic measures is incompatible with the fluid, event-
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
16
based design of the implicit learning measure, the TLT. Thus, innovative future work might investigate the feasibility of discerning the effects of autonomic indices on learning under
PT
different conditions of stimuli frequency (i.e., do autonomic activity effects differ depending on
RI
the familiarity of the event). Future research should also consider developmental aspects of implicit learning and how links between autonomic processes and implicit learning change over
SC
time in order to better understand how decision-making processes develop. Previous research has
NU
demonstrated age differences in implicit learning across adulthood (Mata et al., 2011). However, little is currently known about changes in implicit learning across childhood and adolescence,
MA
except for conclusions inferred from work using tasks that are not entirely implicit, such as the IGT (e.g., Crone et al., 2004). While resting RSA appears to be moderately stable in childhood,
TE
D
RSA reactivity shows less stability (El-Sheikh, 2005). Longitudinal work investigating RSA stability during adolescence to early adulthood and its association with implicit learning could
AC CE P
help to clarify whether RSA plays a role in the dramatic increases in risky decision making and risk-taking behaviors that occur during this often emotionally stressful period.
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
17
References Bechara, A., & Damasio, A.R. (2005). The somatic marker hypothesis: A neural theory of
PT
economic decision. Games and Economic Behavior, 52(2), 336-372.
RI
doi:10.1016/j.geb.2004.06.010
Bechara, A., Damasio, A.R., Tranel, D., & Anderson, S. (1994). Insensitivity to future
SC
consequences following damage to human prefrontal cortex. Cognition, 50(1), 7-15.
NU
doi:10.1016/0010-0277(94)90018-3
Berntson, G.G., Bigger, J.T., Jr., Eckberg, D.L., Grossman, P., Kaufmann, P.G., Malik, M.,
MA
Nagaraja, H.N.,…van der Molen, M.W. (1997). Heart rate variability: Origins, methods, and interpretive caveats. Psychophysiology, 34(6), 623-648.
TE
D
doi:10.1111/j.14698986.1997.tb02140.x Crone, E.A., Somsen, R.J.M., van Beek, B., & van der Molen, M.W. (2004). Heart rate and skin
AC CE P
conductance analysis of antecedents and consequences of decision making. Psychophysiology, 41, 531-540. Crone, E.A., & van der Molen, M.W. (2007). Development of decision making in school-aged children and adolescents: Evidence from heart rate and skin conductance analysis. Child Development, 78(4), 1288-1301. doi:10.1111/j.1467-8624.2007.0106 Damasio, A.R. (1996). The somatic marker hypothesis and the possible functions of the prefrontal cortex. Philosophical Transactions, 351(1346), 1413-1420. doi:10.1098/rstb.1996.0125 Drucaroff, L.J., Kievit, R., Guinjoan, S.M., Gerschcovich, E.R., Cerquetti, D., Leiguarda, R.,…Vigo, D.E. (2011). Higher autonomic activation predicts better performance in Iowa gambling task. Cognitive and Behavioral Neurology, 24(2), 93-98.
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
18
doi:10.1097/WNN.0b013e3182239308 Dunn, B.D., Dalgleish, T., & Lawrence, A.D. (2006). The somatic marker hypothesis: A critical
PT
evaluation. Neuroscience and Biobehavioral Reviews, 30(2), 239-271.
RI
doi:10.1016/j.neubiorev.2005.07.001
Duschek, S., Muckenthaler, M., Werner, N., & Reyes del Paso, G.A. (2009). Relationships
SC
between features of autonomic cardiovascular control and cognitive performance.
NU
Biological Psychology, 81, 110-117. doi:10.1016/j.biopsycho.2009.03.003 El-Sheikh, M. (2005). Stability of respiratory sinus arrhythmia in children and young
MA
adolescents: A longitudinal examination. Developmental Psychobiology, 46(1), 66-74. doi:10.1002/dev.20036
TE
D
Ernst, J.M., Litvack, D.A., Lozano, D.L., Cacioppo, J.T., & Berntson, G.G. (1999). Impedance pneumography: Noise as signal in impedance cardiography. Psychophysiology, 36(3),
AC CE P
333-338. doi:10.1017/S0048577299981003 Fairclough, S.H., Venables, L., & Tattersall, A. (2005). The influence of task demand and learning on the psychophysiological response. International Journal of Psychophysiology, 56, 171-184. doi:10.1016/j.ijpsycho.2004.11.003 Frazier, T.W., Strauss, M.E., & Steinhauer, S.R. (2004). Respiratory sinus arrhythmia as an index of emotional response in young adults. Psychophysiology, 41(1), 75-83. doi:10.1046/j.1469-8986.2003.0013 Frensch, P.A., & Rünger, D. (2003). Implicit Learning. Current Directions in Psychological Science, 12(1), 13-18. doi:10.1111/1467-8721.01213
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
19
Haley, D.W., Grunau, R.E., Oberlander, T.F., & Weinberg, J. (2008). Contingency learning and reactivity in preterm and full-term infants at 3 months. Infancy, 13(6), 570-595.
PT
doi:10.1080/15250000802458682
RI
Hansen, A.L., Johnson, B.H., & Thayer, J.F. (2003). Vagal influence on working memory and attention. International Journal of Psychophysiology, 48, 263-274. doi:10.1016/S0167-
SC
8760(03)00073-4
NU
Hinnant, J.B., El-Sheikh, M., Keiley, M., & Buckhalt, J.A. (2013). Marital conflict, allostatic load, and the development of children’s fluid cognitive performance. Child Development,
MA
84(6), 2003-2014. doi:10.1111/cdev.12103
Howard, J.H., Jr., Howard, D.V., Dennis, N.A., & Kelly, A.J. (2008). Implicit learning of
TE
D
predictive relationships in three-element visual sequences by young and old adults. Journal of Experimental Psychology: Learning, Memory & Cognition, 34(5), 1139-1157.
AC CE P
doi:10.1037/a0012797
Marcovitch, S., Leigh, J., Calkins, S.D., Leerks, E.M., O’Brien, M., & Blankson, A.M. (2010). Moderate vagal withdrawal in 3.5-year-old children is associated with optimal performance on executive function tasks. Developmental Psychobiology, 52(6), 603-608. doi:10.1002/dev.20462
Mata, R., Josef, A.K., Samanez-Larkin, G.R., & Hertwig, R. (2011). Age differences in risky choice: A meta-analysis. Annals of the NY Academy of Sciences, 1235(1), 18-29. doi:10.1111/j.1749-6632.2011.0620 Overbeek, T.J.M., van Boxtel, T., & Westerink J.H.D.M. (2014). Respiratory sinus arrhythmia responses to cognitive tasks: Effects of task factors and RSA indices. Biological Psychology, 99, 1-14. doi:10.1016/j.biopsycho.2014.02.006
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
20
Pascual-Leone, A., Wassermann, E.M., Grafman, J., & Hallett, M. (1996). The role of the dorsolateral prefrontal cortex in implicit procedural learning. Experimental Brain
PT
Research, 107(3), 479-485.
RI
Porges, S.W. (1992). Vagal tone: A physiological marker of stress vulnerability. Pediatrics, 90(3), 498-504.
SC
Porges, S.W. (2003). The polyvagal theory: phylogenetic contributions to social behavior.
NU
Physiology & Behavior, 79(3), 503-513. doi:10.1016/S0167-8760(01)00162-3 Porges, S.W. (2007). The polyvagal perspective. Biological Psychology, 74(2), 116-143.
MA
doi:10.1016/j.biopsycho.2006.06.009
Reber, A.S. (1989). Implicit learning and tacit knowledge. Journal of Experimental Psychology:
TE
D
General, 118(3), 219-235. doi:10.1037/0096-3445.118.3.219 Simon, J.R., Howard, J.H., Jr., & Howard, D.V. (2010). Adult age differences in learning from
AC CE P
positive and negative probabilistic feedback. Neuropsychology, 24(4), 534-541. doi:10.1037/a0018652
Simon, J.R., Vaidya, C.J., Howard, J.H., Jr., & Howard, D.V. (2012). The effects of aging on the neural basis of implicit associative learning in a probabilistic triplets learning task. Journal of Cognitive Neuroscience, 24(2), 451-463. doi:10.1162/jocn_a_00116 Staton, L., El-Sheikh, M., & Buckhalt, J.A. (2009). Respiratory sinus arrhythmia and cognitive functioning in children. Developmental Psychobiology, 51(3), 249-258. doi:10.1002/dev.20361 Stillman, C.M., Feldman, H., Wambach, C.G., Howard, J.H., Jr., & Howard, D.V. (2014). Dispositional mindfulness is associated with reduced implicit learning. Consciousness and Cognition, 28, 141-150. doi:10.1016/j.concog.2014.07.002
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
21
Thayer, J.F., Ahs, F., Fredrikson, M., Sollers, J.J., III. & Wager, T.D. (2012). A meta-analysis of heart rate variability and neuroimaging studies: Implications for heart rate variability as a
PT
marker of stress and health. Neuroscience and Biobehavioral Reviews, 36, 747-756.
RI
doi:10.1016/j.neubiorev.2011.11.009
Thayer, J.F., & Lane, R.D. (2000). A model of neurovisceral integration in emotion regulation
SC
and dysregulation. Journal of Affective Disorders, 61, 201-216.
NU
Yang, J., & Li, P. (2012). Brain networks of explicit and implicit learning. PLoS One, 7(8),
AC CE P
TE
D
MA
e42993. doi:10.1371/journal.pone.0042993
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
22
Cue 1
PT
Cue 2
RI
Target
SC
270 ms
NU
270 ms
MA
Respond
D
Figure 1. An example of a trial on the TLT. Participants observe the first two events and
AC CE P
TE
respond only to the location of the target event.
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING 460
440
PT
430 420 410
RI
High Probability Low Probability
SC
400 390 380 370 360 2
3 Epoch
MA
1
NU
Mean Median RT (in msec)
450
4
D
Figure 2. Mean of median RT (in milliseconds) over epochs by high- and low-probability
AC CE P
TE
triplets. Error bars represent the standard error of the mean.
23
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
25
PT
High Resting RSA; Less RSA withdrawal
15
High Resting RSA; Greater RSA withdrawal
5 0 Epoch 1
Epoch 2
Epoch 3
Epoch 4
SC
RI
10
NU
Triplet Type Effect (in ms)
20
Low Resting RSA; Greater RSA withdrawal
MA
-5
Low Resting RSA; Less RSA withdrawal
-10
TE
D
Figure 3. The interaction of resting RSA and RSA reactivity on learning across the TLT. A
AC CE P
larger triplet type effect (TTE) indicates greater learning.
24
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING
RI
PT
Table 1. Descriptive statistics for study variables. Variable M SD Resting RSA 6.10 1.05 RSA Reactivity -0.06 0.60 TTE Epoch 1 -0.89 28.29 TTE Epoch 2 10.92 18.82 TTE Epoch 3 8.75 18.34 TTE Epoch 4 16.37 20.99
25
SC
Note. RSA Reactivity was calculated as the difference between average RSA during the
NU
Triplets Learning Task (TLT) and resting RSA. Therefore, a positive reactivity value denotes an increase in RSA from baseline to the task. TTE was calculated as the average difference in
MA
reaction time to low- and high-probability triplets. Thus, a positive TTE value indicates, on average, faster responding to high- relative to low-probability triplets. RSA = respiratory sinus
AC CE P
TE
D
arrhythmia. TTE = triplet type effect.
ACCEPTED MANUSCRIPT Running head: AUTONOMIC ACTIVITY AND IMPLICIT LEARNING Highlights
TE
D
MA
NU
SC
RI
PT
Respiratory sinus arrhythmia (RSA) predicts change over time in implicit learning Greater resting RSA and RSA withdrawal predicts better implicit learning Links between RSA and learning may help explain decision making under uncertainty
AC CE P
26