Goal pursuit is among the most fundamental of human psychological activities (Miller, Galanter, & Pribram, Reference Miller, Galanter and Pribram1960). Much of human behavior can be understood to reflect people’s efforts to pursue desired outcomes and avoid undesired ones (Deci & Ryan, Reference Deci and Ryan2000). Different types of goals are associated with different motivational states, means of pursuit, and affective/behavioral consequences. The types of goals people pursue and the ways in which they pursue them have been used to account for a broad spectrum of behavioral differences as well as emotional vulnerabilities such as depression (Strauman, Reference Strauman2017). In order for effective goal pursuit to occur, the individual must have the capacity for self-regulation (Hoyle & Gallagher, Reference Hoyle, Gallagher, Mikulincer, Shaver, Cooper and Larsen2015), particularly including the processes by which individuals identify and pursue goals and assess their progress toward them (Carver & Scheier, Reference Carver and Scheier1990). Effective self-regulation is supported by executive functioning skills such as attention, working memory, response inhibition, and task switching (Hofmann, Schmeichel, & Baddeley, Reference Hofmann, Schmeichel and Baddeley2012; Hofmann, Schmeichel, Friese, & Baddeley, Reference Hofmann, Schmeichel, Friese, Baddeley, Vohs and Baumeister2010), which operate together to create the capacity to represent and pursue higher order goals, initiate and sustain adaptive motivational orientations, and respond to goal-relevant feedback (Baumeister & Heatherton, Reference Baumeister and Heatherton1996; Carver & Scheier, Reference Carver and Scheier1981; Hofmann et al., Reference Hofmann, Schmeichel and Baddeley2012).

Personal goal pursuit has been studied extensively using regulatory focus theory (RFT; Higgins, Reference Higgins1997, Reference Higgins1998). RFT postulates two cognitive-motivational systems that correspond to distinct self-regulatory orientations for attaining desired end states: prevention and promotion. A prevention orientation emphasizes duties, obligations, and responsibilities, which in turn are related to fundamental needs for security. Goals are viewed as oughts, and there is a strategic concern with, n lay person’s terms, “keeping bad things from happening.” A promotion orientation emphasizes hopes, accomplishments, and advancement, which in turn are related to fundamental needs for nurturance. Goals are viewed as ideals, and there is a strategic emphasis on “making good things happen.” Regulatory focus is a psychological state that varies both across individuals (when construed as a dispositional variable) and within individuals (on a momentary basis across situations). The promotion and prevention systems are postulated to be functionally distinct from the behavioral activation system (BAS; Depue & Iacono, Reference Depue and Iacono1989; Gray, Reference Gray1990) and behavioral inhibition system (BIS; Fowles, Reference Fowles1988; Gray, Reference Gray1982). For example, while both promotion/prevention and BIS/BAS can be conceptualized broadly in terms of approach and avoidance (Scholer & Higgins, Reference Scholer, Higgins and Elliot2008), RFT emphasizes strategic pursuit of higher order goals whereas BIS/BAS refers primarily to spatiotemporal, evolutionarily determined approach/avoidance behaviors (Depue & Collins, Reference Depue and Collins1999; Strauman & Wilson, Reference Strauman, Wilson and Hoyle2010).

RFT predicts how perceived successes or failures in goal pursuit can give rise to different emotions. Self-discrepancy theory (SDT; Higgins, Reference Higgins1987), a precursor of RFT, proposed that a perceived discrepancy between one’s actual self and one’s ideal versus ought self, which represent motivationally significant self-guides, creates different emotional responses. Attaining or not attaining a particular self-guide was postulated to represent a specific psychological situation (Lewin, Reference Lewin and Cartwright1951), and self-regulation in reference to ideal and ought self-guides illustrates the motivational and affective distinctions between promotion and prevention goal pursuit. According to SDT, believing that one is attaining an ideal (promotion) goal is associated with emotions such as joy, satisfaction, and happiness, whereas believing that one is not attaining such a goal is associated with sadness, disappointment, and frustration. Likewise, believing that one is attaining an ought (prevention) goal is associated with emotions such as quiescence and calmness, whereas believing that one is not attaining such a goal is associated with anxiety, worry, and agitation.

A number of investigators have sought to identify neural correlates of promotion and prevention (e.g., Cunningham et al., Reference Cunningham, Raye and Johnson2005; Packer & Cunningham, Reference Packer and Cunningham2009; Touryan et al., Reference Touryan, Johnson, Mitchell, Farb, Cunningham and Raye2007). The logic for such an investigation reflects the hypothesized development of promotion and prevention orientations as derived from the individual’s socialization history (Scholer & Higgins, Reference Scholer, Higgins and Elliot2008). Specifically, strength of orientation to promotion versus prevention is postulated to be determined by experienced patterns of caretaker/child interactions (Higgins, Reference Higgins1998; Manian, Papadakis, Strauman, & Essex, Reference Manian, Papadakis, Strauman and Essex2006). Therefore, individuals would be expected to vary in the strength of neural responses to cues for promotion versus prevention goal pursuit, just as they have been observed to vary with regard to motivational, affective, and behavioral consequences of promotion or prevention goal activation. The initial studies linking regulatory focus with brain activation drew extensively on the broader literature on frontal asymmetry and positive versus negative affectivity (e.g., Coan & Allen, Reference Coan and Allen2003). Amodio, Shah, Sigelman, Brazy, and Harmon-Jones (Reference Amodio, Shah, Sigelman, Brazy and Harmon-Jones2004) examined the associations between individual differences in chronic regulatory focus and an EEG measure of resting prefrontal cortex (PFC) activation asymmetry. Individuals with a chronic promotion focus manifested higher levels of resting left PFC activity, whereas those with a chronic prevention focus manifested greater right PFC activity. Eddington et al. (Reference Eddington, Dolcos, Cabeza, Krishnan and Strauman2007) investigated cortical activation patterns associated with promotion versus prevention goals, using a functional magnetic resonance imaging (fMRI) idiographic priming paradigm in which participants’ promotion and prevention goals were presented incidentally within a depth-of-processing task. Priming with promotion goals was associated with left PFC activation, and the observed activation was stronger for individuals with a chronic promotion focus. In a subsequent study, Eddington et al. (Reference Eddington, Dolcos, McLean, Krishnan, Cabeza and Strauman2009) found the same discriminant association between promotion goal priming and left PFC activation and observed that depressed individuals manifested a significantly attenuated LPFC response when primed with their own promotion goals. More recently, Shi, Ma, Wu, Wu, Wang, and Han (Reference Shi, Ma, Wu, Wu, Wang and Han2016) found that larger actual:ideal self-discrepancy was associated with activations in the ventral/dorsal striatum and dorsomedial and lateral prefrontal cortices (PFCs). Similarly, Scult et al. (Reference Scult, Knodt, Hanson, Ryoo, Adcock, Hariri and Strauman2017) used a monetary incentive delay task to activate reward circuitry and observed that promotion orientation was inversely correlated with ventral striatum response to gain cues.

There are a number of well-articulated models that postulate brain systems that underlie or instantiate important self-regulatory functions (e.g., Badre & Nee, Reference Badre and Nee2018), including several that address personal goal pursuit. Heatherton (Reference Heatherton2011) proposed a conceptual framework for interpreting associations between different aspects of self-regulation and activity in different brain regions. This framework was elaborated by Kelley, Wagner, and Heatherton (Reference Kelley, Wagner and Heatherton2015) and provides a comprehensive picture of functionally connected brain regions that are engaged under different self-regulatory circumstances. For example, contemplation of personal goals has been observed to elicit activation in brain regions linked to self-reflection and/or encompassed by the default mode network (Spreng, Stevens, Chamberlain, Gilmore, & Schacter, Reference Spreng, Stevens, Chamberlain, Gilmore and Schacter2010). The default mode network, which includes the medial PFC, posterior cingulate cortex, precuneus, lateral and medial temporal lobes, and posterior inferior parietal lobule, is a network of brain regions thought to be preferentially active during internally focused cognition (Buckner, Andrews-Hanna, & Schacter, Reference Buckner, Andrews-Hanna and Schacter2008). Activation of regions associated with the default mode network may thus be observed in response to priming of salient promotion or prevention goals, due to the self-referential cognition required in order to define, maintain, and regulate toward such goals.

In addition to regions included in the default mode network, there is a partially overlapping network of regions along the cortical midline that has been shown to be critical for self-referential thought and therefore likely to be engaged during the process of personal goal pursuit (Northoff & Bermpohl, Reference Northoff and Bermpohl2004; Qin & Northoff, Reference Qin and Northoff2011). This cortical midline structures model includes the orbital and dorsomedial PFC, the anterior cingulate cortex, and the posterior cingulate cortex including the precuneus (Northoff & Bermpohl, Reference Northoff and Bermpohl2004). These structures are associated with the representation and processing of self-relevant stimuli (Beer & Ochsner, Reference Beer and Ochsner2006; Northoff & Bermpohl, Reference Northoff and Bermpohl2004), suggesting their potential relevance for identifying broad patterns of activation following promotion/prevention priming.

Most recently, Langner, Leiberg, Hoffstaedter, and Eickhoff (Reference Langner, Leiberg, Hoffstaedter and Eickhoff2018) conducted two meta-analyses of human neuroimaging studies on emotion or action control. Their conceptual framework emerged in part from a study by Murray, Debbane, Fox, Bzdok, and Eickhoff (Reference Murray, Debbané, Fox, Bzdok and Eickhoff2015) that combined task-dependent meta-analytic connectivity modeling with task-independent resting-state connectivity analysis and focused on similarities and distinctions between self-processing and other processing. Murray and colleagues found two functionally connected subnetworks, one preferentially activated during self-relevant processing and the other during other-relevant processing. Langer and colleagues summarized a partially distinct set of studies and reported two fronto-parieto-insular networks with limited overlap (bilateral anterior insula, posteromedial PFC, and right temporo-parietal junction). The investigators went on to postulate a core brain/behavior system for implementing self-control across both emotion and goal-directed action but also noted that there were distinctions on functional connectivity analyses between components of the system related to affective versus executive aspects of self-control.

To date, several neuroimaging studies of promotion and prevention focus have found activation in one or more of the aforementioned self-regulation-relevant regions. Johnson et al. (Reference Johnson, Raye, Mitchell, Touryan, Greene and Nolen-Hoeksema2006) used fMRI to investigate differences in neural activation in response to promotion and prevention goals in contrast with nonself-relevant items. Their results indicated that reflecting on promotion/prevention goals was associated with greater activity in both the medial/anterior cingulate cortex and the posterior cingulate cortex/precuneus. Moreover, they found that the frontal regions were relatively more active in response to reflecting on promotion goals, whereas the posterior regions were more active in reflecting on prevention goals. Packer and Cunningham (Reference Packer and Cunningham2009) investigated whether temporal proximity to a promotion or prevention goal altered neural activation. In addition to observing differences based on how near or far a hypothetical goal was in time, they again observed activation in the medial PFC, anterior cingulate cortex, and the precuneus; however, they did not observe the same dissociation between promotion and prevention goals as was reported by Johnson et al. (Reference Johnson, Raye, Mitchell, Touryan, Greene and Nolen-Hoeksema2006). As noted above, Shi et al. (Reference Shi, Ma, Wu, Wu, Wang and Han2016) found that actual:ideal self-discrepancy predicted activation in the dorsal medial and lateral PFCs.

Strauman et al. (Reference Strauman, Detloff, Sestokas, Smith, Goetz, Rivera and Kwapil2013) used a different task-based fMRI approach to further elucidate the neural correlates of promotion and prevention goal pursuit. Applying a rapid masked priming technique in which participants were exposed subliminally to their own previously assessed promotion and prevention goals, the authors observed distinct patterns of neural activation associated with promotion versus prevention goals. Promotion priming led to activation in frontal and occipital regions as well as caudate and thalamus, whereas prevention priming was associated with activation in precuneus and posterior cingulate cortex. Individual differences in chronic dysphoric/anxious affect and in regulatory focus predicted differential activation following promotion versus prevention priming. The regions activated in response to promotion and prevention goals mapped broadly onto the cortical midline structures network. However, Strauman et al. (Reference Strauman, Detloff, Sestokas, Smith, Goetz, Rivera and Kwapil2013) did not systematically vary the extent to which the goals being used as priming stimuli were perceived by participants as cues for success (i.e., goals they believed they were attaining or had attained) versus failure (i.e., goals they believed they had not attained). According to RFT, this dimension of goal representation and goal pursuit processing should have profound motivational and affective consequences, which in turn could be associated with distinct neural activation patterns.

To our knowledge, no study has systematically manipulated both dimensions that SDT and RFT hypothesize as critical for determining the motivational and affective significance of goal activation: domain of goal (i.e., promotion/ideal vs. prevention/ought) and attainment status (i.e., one’s perceived behavior/attributes matching versus mismatching a motivationally significant goal). In the present study, we asked two related questions: first, what patterns of neural activity are associated with priming of promotion versus prevention goals also considering the perceived attainment status of each goal; and second, to what extent are the observed patterns of neural activity for each type of goal related to a self-report index of individual differences in chronic regulatory focus.

We chose to examine the data using two complementary analytic approaches: a standard general linear model (GLM) whole-brain analysis looking for statistically significant regions of activation associated with a particular theory-based contrast (e.g., promotion match goal priming vs. control priming) and a task-based functional connectivity analysis looking for common versus unique patterns of activation across the domain of goal and attainment status priming manipulations. The standard GLM analysis allowed us to compare our findings directly with published studies in a larger sample, whereas the functional connectivity analysis was intended to examine the overlap between neural activations following promotion versus prevention goal priming and existing cognitive neuroscience models of self-regulation. Following the recommendations of Beckmann and Smith (Reference Beckmann and Smith2005) and Guo and Pagnoni (Reference Guo and Pagnoni2008), tensorial probabilistic independent component analysis (TICA) was chosen for the task-based functional connectivity analysis because it affords two primary benefits when compared with seed-based psychophysiological interaction analysis. First, it identifies a set of statistically independent components, a subset of which can then be selected for further examination based on a priori criteria. This process reduces the number of comparisons and increases statistical power to detect differences across priming conditions. Second, because TICA extracts components without a model or spatial parameters and then tests for statistically significant associations with task-based model parameters, it provides an optimal blend of exploratory and hypothesis-testing approaches to functional connectivity – particularly valuable in this context because of the likelihood that activation across the priming conditions would have both common and unique components. In the following sections, we describe the methods and findings following the reporting recommendations of Nichols et al. (Reference Nichols, Das, Eickhoff, Evans, Glatard, Hanke and Yeo2016, Reference Nichols, Das, Eickhoff, Evans, Glatard, Hanke and Yeo2017).

1. Method

1.1 Participants

A total of 55 participants were approached through two sources (see below), and all 55 responded in the affirmative and then gave informed consent and completed the study. A subset of participants (N = 7) was excluded from analyses due to either excessive head motion during scanning (≥3.0 mm; N = 2), exceeding threshold on one or more quality control measures (see below; N = 3), or technical issues while scanning (N = 2), leaving a final sample of 48 participants with complete data (28 females and 20 males). Participants had a mean age of 28.32 ± 11 years (range = 18–60), and all were right-handed as per self-report. Of these 48 participants, 32 were Caucasian, 6 were African-American, and 10 were Asian. Participants were recruited either through the participant pool at the Duke Interdisciplinary Institute for Social Psychology Lab or through the Duke Neurogenetics Study, a longitudinal investigation in which participants had agreed to be recontacted for further research participation. Informed consent was obtained prior to participation. Participants were compensated $20 per hour.

1.2 Procedure

The study consisted of two sessions conducted approximately 4 weeks apart. In the first session, participants completed the Regulatory Focus Questionnaire (RFQ; Higgins et al., Reference Higgins, Friedman, Harlow, Idson, Ayduk and Taylor2001) and the Computerized Goal Assessment task (Shah & Higgins, Reference Shah and Higgins2001), which provided the data needed to create the priming stimuli and the measures of individual differences in promotion and prevention orientation. In the second session, participants were scanned while completing an idiographic subliminal priming task, debriefed, and compensated for their time. All procedures were approved by the Duke University Institutional Review Board.

The RFQ is a self-report measure designed to assess the degree to which individuals are oriented toward promotion and/or prevention. The RFQ used in this study contained 12 Likert style items divided into two subscales measuring the extent to which an individual believes they have been successful in attaining promotion or prevention goals. The subscales include items such as: “I feel like I have made progress toward being successful in my life” (promotion success) and “Not being careful enough has gotten me in trouble at times” (prevention success, reverse-scored). Each subscale has an internal consistency (coefficient alpha) of .75 or higher and a 2-month test–retest reliability (Pearson correlation) of .79 or higher.

The Computerized Goal Assessment task (Shah & Higgins, Reference Shah and Higgins2001) is a computer-administered measure of self-reported attributes that describe one’s ideal and ought selves. Participants were first given definitions of what ideal and ought attributes are (i.e., “attributes of the kind of person you wish or desire to be” and “attributes of the person you believe it is you duty or responsibility to be,” respectively). They were then serially prompted to think of six attributes to describe their ideal self and six attributes to describe their ought self, in alternating order. Participants were discouraged from listing more than one word for each response, from repeating responses, or from listing synonyms. After each response, participants were then asked to rate the attribute using a scale from one (not at all) to seven (extremely) for the following question: “How far are you from possessing this attribute?”. Four attributes were selected from each category (i.e., ideal and ought) for each participant to be used as self-relevant priming stimuli in the subsequent scanning session: the two most highly “matched” attributes (i.e., the attributes they rated as being closest to possessing on the 7-point scale) and the two most highly “mismatched” attributes (i.e., the attributes they rated as being farthest from possessing on the 7-point scale). Four yoked-control attributes were also selected from the responses of another participant selected at random (two ideal adjectives and two ought adjectives); these words were semantically unrelated to the attributes listed by the participant.

In a subsequent session, participants then completed an event-related rapid masked priming fMRI task (Strauman et al., Reference Strauman, Detloff, Sestokas, Smith, Goetz, Rivera and Kwapil2013; see Figure 1). In this task, modeled on the task developed by Diaz and McCarthy (Reference Diaz and McCarthy2007), participants were exposed to a continuous series of rapidly presented, masked visual stimuli. These stimuli included the participants’ previously listed ideal and ought attributes, the ideal and ought attributes of another participant as a yoked-control comparison condition, and nonword letter strings. All goal words and all yoked-control words were positively valenced trait attributes; nonwords were random consonant strings, each 4–10 characters in length. Each letter string was padded with pound signs so that letters were centered and each stimulus was 12 characters in length to ensure that the same amount of the visual field was occupied at any given time during the experiment.

Figure 1.

Schematic of the experimental task. The sequence for an individual trial consisted of alternating pound signs and percent signs, in between which a word or nonword was inserted. Incidental to the promotion goal, prevention goal, and yoked-control priming stimuli that were inserted throughout the run, visible colored stimuli were displayed to which participants were instructed to press a button as quickly as possible.

A critical stimulus occurred approximately every 500 ms within the constantly changing visual display; the duration of critical stimulus presentation was 33 ms. Trials primarily consisted of masked nonwords, and masked goal words or yoked-control words were presented once every 12–15 s. All word and nonword strings were preceded and followed by longer duration pound sign strings, which served as pattern masks. The pound sign strings in turn alternated with percent sign strings such that participants experienced a continuously changing visual stream. As an attentional control, participants were instructed to make a button-press response when they detected a pound sign string presented in red font. These target events occurred infrequently (mean interval = 25 s) and for a longer duration (155 ms) and were not in close temporal proximity to the masked attribute trials. Participants were not told that either words or nonword strings would be presented.

The task was divided into four runs of 6 min each. The run conditions were as follows: ideal match (ideal attributes rated as closest to possessing), ideal mismatch (ideal attributes rated as farthest from possessing), ought match (ought attributes rated as closest to possessing), and ought mismatch (ought attributes rated as farthest from possessing). Each run also included yoked-control attributes (goal attributes selected from a different subject that were not semantically related to any of the target participant’s goals) to allow for creation of contrasts. Run order was randomized across participants to control for any order effects, and initial statistical analyses detected no statistically significant effect of run order. Stimuli were presented using CIGAL, an in-house data presentation software program (Voyvodic, Reference Voyvodic1999). All stimuli were displayed on MRI-compatible LCD goggles.

1.2.1 Image acquisition

Neuroimaging data were acquired on a 3.0 Tesla GE Signa EXCITE HD system (Waukesha, WI, USA) at Duke’s Brain Imaging and Analysis Center (www.biac.duke.edu). Four initial RF excitations were acquired and discarded to achieve steady-state equilibrium. For each subject, 720 functional images (180 per run) with 34 interleaved 4-mm-thick slices were acquired in the oblique axial plane using a T2* -weighted sense spiral sequence (TR/TE = 2000/30 ms, voxel dimensions = 3.8 × 3.8 × 3.8 mm, flip angle = 60°, field of view = 240 mm, matrix size = 64 × 64). Prior to functional image collection, a T1-weighted coplanar structural scan with 34.4-mm slices (TR/TE = 7600/30 ms, flip angle = 12°; field of view = 240 mm, matrix size = 256 × 256) was acquired parallel to the AC–PC line for spatial normalization and coregistration of the functional data to a standard atlas.

1.2.2 Quality control

Prior to preprocessing the fMRI data, we obtained three measures of data quality and excluded subjects with extreme values. First, we estimated the average signal-to-fluctuation-noise ratio for each subject, defined as the mean of the signal across time divided by the standard deviation of the signal across time. Second, we computed the mean volume-to-volume head motion (i.e., displacements relative to the preceding time point in units of millimeter) for each subject. Third, using an FMRIB’s Software Library (FSL) tool called fsl_motion_outliers, we identified outlier volumes in the functional data by examining the root-mean-square error (RMSE) of each volume relative to the reference volume (with the middle time point used as reference). A volume was judged to be an outlier if its RMSE amplitude exceeded the 75th percentile plus the value of 150% of the interquartile range of RMSE for all volumes in a run (representing a standard boxplot threshold). We excluded subjects where any of the three quality control measures was extreme (beyond the upper or lower 5th percentile). This procedure resulted in excluding the data from three participants.

1.2.3 fMRI data preprocessing

The functional data were preprocessed using FMRI Expert Analysis Tool within FSL (Oxford University; www.fmrib.ox.ac.uk/fsl). The following steps were utilized: motion correction by coregistration of all volumes to a reference volume using MCFLIRT nonbrain structures removal using Brain Extraction Tool spatial smoothing using a Gaussian kernel of 8.0-mm Full Width Half Maximum; grand mean intensity normalization of the entire 4D dataset by a single multiplicative factor; and high-pass temporal filtering (Gaussian-weighted least squares straight line fitting, 50-s cutoff).

1.2.4 GLM analyses

Whole-brain GLM-based analyses were conducted to identify regions activated by promotion and prevention goal priming for both match and mismatch conditions. The GLM analysis methods closely followed past procedures (Strauman et al., Reference Strauman, Detloff, Sestokas, Smith, Goetz, Rivera and Kwapil2013) in order to determine whether previous findings were replicated in the present sample. All statistical images were thresholded using clusters determined by Z > 2.3 and a whole-brain corrected cluster significance threshold of p < .05, supplemented by the nonparametric permutation method recommended by Eklund, Nichols, & Knutson (Reference Eklund, Nichols and Knutson2016). First, denoized functional data were analyzed using a GLM with local autocorrelation correction (Woolrich, Ripley, Brady, & Smith, Reference Woolrich, Ripley, Brady and Smith2001). For each run, separate regressors were created for the goal priming words, the yoked-control words, and the nonword stimuli. A nuisance regressor modeled the button-press response component of the task. All regressors consisted of unit impulses convolved with a canonical hemodynamic response function. Four contrasts were created for each participant: (1) ideal match versus yoked-control, (2) ideal mismatch versus yoked-control, (3) ought match versus yoked-control, and (4) ought mismatch versus yoked-control. Data were combined across subjects using a mixed-effects model (Beckmann, Jenkinson, & Smith, Reference Beckmann, Jenkinson and Smith2003; Woolrich, Behrens, Beckmann, Jenkinson, & Smith, Reference Woolrich, Behrens, Beckmann, Jenkinson and Smith2004). Mixed-effects models were then used to obtain statistical tests for whether the two individual difference measures (promotion and prevention success) were associated with activation following goal priming (again using the four contrasts created for each participant). Analyses of the priming condition contrasts and the individual differences measures were carried out using FLAME (FMRIB’s Local Analysis of Mixed Effects) – stage 1 (Beckmann et al., Reference Beckmann, Jenkinson and Smith2003; Woolrich et al., Reference Woolrich, Behrens, Beckmann, Jenkinson and Smith2004), consisting on a mixed-effects (fixed and random effects) approach using Bayesian estimation techniques. To account for the likelihood that some brain regions might be responsive to particular kind of priming stimuli only among individuals characterized by high or low levels of chronic perceived success in personal goal pursuit, the individual difference measures were included in the whole-brain analyses. Throughout the analyses, no statistically significant main effects or interactions involving participant gender or run order were identified. Therefore, none of the analyses reported below include gender or run order as a between-subject effect.

1.2.5 Independent component analyses

TICA is a technique that extends single-session independent component analysis (ICA) to higher dimensions as an approach to multisession/multisubject task-based fMRI data analysis. Our TICA analysis was performed with Multivariate Exploratory Linear Decomposition into Independent Components (MELODIC) Version 3.14 from the FSL (www.fmrib.ox.ac.uk/fsl). Four TICA analyses were conducted, one for each priming condition contrast corresponding to each of the GLM analyses for priming condition. Within each analysis, all subjects were time concatenated for a single 48-run (48 subjects) group ICA, using the same preprocessed functional data that were entered into the GLM analyses. All subjects’ data were spatially normalized to an anatomical MNI standard template using a 12-parameter affine registration implemented in FLIRT). The voxel BOLD times series were demeaned, variance normalized, and whitened. The number of independent components extracted by MELODIC was estimated using a Bayesian approach. Using Probabilistic Independent Component Analysis (Beckmann & Smith, Reference Beckmann and Smith2005), the whitened time data were projected onto a multidimensional subspace. Fixed-point iteration optimization was used to decompose the projected data into independent vector sets that account for variability in temporal, spatial, and subject domains. The spatial components were normalized by the variance of the residuals, and a mixture model was fit to the obtained intensity histograms to determine a statistical threshold (Beckmann & Smith, 2005).

Within each TICA analysis, we included a regressor for the contrast of interest (e.g., ideal/match vs. yoked-control) as well as regressors for the RFQ promotion and prevention scores. For each analysis, we used the following a priori criteria to determine the number of components for further examination (following the recommendations of Najeed & Avison, Reference Najeed and Avison2014): (1) the component needed to account for at least 5.0% of the total variance within that analysis, (2) the component did not resemble artifact or noise upon visual inspection, particularly with reference to the timing of activation peaks in comparison with the contrast regressors, (3) the overall F value for the set of three theory-based regressors (contrast of interest, RFQ promotion score, and RFQ prevention score) needed to meet statistical significance at p < .01, and (4) at least one of the three regressors needed to meet statistical significance at p < .01. These more lenient statistical criteria were used in order to maximize the exploratory power of the TICA analyses, given the more conservative nature of this analytic method compared with GLM-based analyses (Beckmann & Smith, Reference Beckmann and Smith2005).

1.2.6 Statistical power

Power analysis was performed following the recommendations of Geuter et al. (Reference Geuter, Qi, Welsh, Wager and Lindquist2018) and Mumford (Reference Mumford2012). Power estimation was based on the findings of Strauman et al. (Reference Strauman, Detloff, Sestokas, Smith, Goetz, Rivera and Kwapil2013) who used the identical idiographic goal priming procedure and conducted whole-brain GLM analyses with priming condition/contrast, a within-subject manipulation, as the primary predictor of interest. Strauman et al. observed a medium-to-large effect size (Cohen’s d = .68) based on a null-hypothesis test of the ideal versus control priming contrast of parameter estimates. Based on those data, the present study (with a final N of 48 participants) had 71% power to detect a similar sized effect in whole-brain GLM analyses.

2. Results

2.1 GLM analyses: Activation in response to each goal priming condition

In order to identify activation in response to being primed by matching or mismatching promotion or prevention goals, BOLD signal change in response to each type of priming word was determined contrasted with activation in response to yoked-control words within the same run. To account for the likelihood that some regions might be responsive to particular kinds of priming stimuli only among individuals characterized by high or low levels of chronic perceived success in personal goal pursuit, the promotion and prevention success scores were included in these analyses as covariates.

2.1.1 Ideal match versus yoked-control contrast

Table 1 lists the regions showing statistically significant activation for the ideal match versus control contrast. For the contrast main effect, there was a single cluster comprised of regions in bilateral parietal lobe (primarily bilateral precuneus, Brodmann area (BA) 7). In addition, scores on the promotion success subscale were significantly positively correlated with activation on the ideal match versus control contrast in two distinct clusters. The first cluster was distributed across the frontal, temporal, and occipital lobes including left middle frontal gyrus (BA 9), left inferior and superior temporal gyrus (BA 20 and 38), and left fusiform gyrus (BA 10). The second cluster was distributed across the temporal, occipital, and parietal lobes and included right inferior and middle temporal gyrus and precuneus (BA 7, 20, and 21) and right lateral occipital cortex (BA 19).

Table 1.

Regions associated with the ideal/match versus yoked-control contrast in GLM analysis

Regions manifesting statistically significant activation in the ideal/match versus yoked-control contrast. N = 48. Coordinates of the top six local maxima within each cluster of activation are in MNI space. Probabilistic labels reflect the likelihood that a coordinate belongs to a given region; only the label with the highest probability is shown.

2.1.2 Ideal mismatch versus yoked-control contrast

Table 2 lists the regions showing statistically significant activation for the ideal mismatch versus control contrast. For contrast main effect, there was a single cluster comprised of regions in bilateral occipital lobe (primarily bilateral cuneus, BA 18). In addition, scores on the promotion success subscale were significantly associated (and positively correlated) with activation on the ideal mismatch versus control contrast in a single cluster which was centered in the left frontal lobe (specifically left middle frontal gyrus and inferior frontal gyrus, BA 9 and 45/46/47, respectively).

Table 2.

Regions associated with the ideal/mismatch versus yoked-control contrast in GLM analysis

Regions manifesting statistically significant activation in the ideal/mismatch versus yoked-control contrast. N = 48. Coordinates of the top six local maxima within each cluster of activation are in MNI space. Probabilistic labels reflect the likelihood that a coordinate belongs to a given region; only the label with the highest probability is shown.

2.1.3 Ought match versus yoked-control contrast

Table 3 lists the regions showing statistically significant activation for the ought match versus control contrast. For the contrast main effect, two significant clusters were found. The first cluster was distributed across the frontal lobes and included right middle and superior frontal gyrus (BA 8/9/10) and left middle frontal gyrus (BA 6). The second cluster was found in the parietal lobes and included right and left supramarginal gyrus and right angular gyrus (BA 7 and 40). There were no regions of activation for the ought match versus control contrast which were significantly correlated with scores on either the promotion or prevention success subscales.

Table 3.

Regions associated with the ought/match versus yoked-control contrast in GLM analysis

Regions manifesting statistically significant activation in the ought/match versus yoked-control contrast. N = 48. Coordinates of the top six local maxima within each cluster of activation are in MNI space. Probabilistic labels reflect the likelihood that a coordinate belongs to a given region; only the label with the highest probability is shown.

2.1.4 Ought mismatch versus yoked-control contrast

Table 4 lists the regions showing statistically significant activation for the ought mismatch versus control contrast. For the contrast main effect, one significant cluster was found. The cluster included maxima in bilateral caudate as well as bilateral occipital lobe, specifically left and right intracalcarine cortex (BA 18). There were no regions of activation for the ought match versus control contrast which were significantly correlated with scores on either the promotion or prevention success subscales.

Table 4.

Regions associated with the ought/mismatch versus yoked-control contrast in GLM analysis

Regions manifesting statistically significant activation in the ought/mismatch versus yoked-control contrast. N = 48. Coordinates of the top six local maxima within a single cluster of activation are in MNI space. Probabilistic labels reflect the likelihood that a coordinate belongs to a given region; only the label with the highest probability is shown.

2.2 TICA analyses: Independent components associated with priming condition and/or RFQ scores

For each of the four separate TICA analyses (each corresponding to one of the four GLM analyses by priming condition), we applied the component selection criteria described above to identify components that accounted for sufficient variance and were associated with either the contrast of interest per se or to the RFQ promotion or prevention scores.

2.2.1 Components associated with the ideal match versus yoked-control contrast

We found three components of interest related to this priming condition. The first component accounted for 6.91% of total variance, had an excellent overall model fit, F(3, 45) = 26.00, p < .0001, and was significantly associated with both the contrast per se and with scores on the promotion success scale, Z = 6.58 (p < .001) and 2.22 (p < .02), respectively. The loci of greatest activation were in bilateral insula (BA 13) and bilateral precentral gyrus (BA 44), with no regions showing significant deactivation. The second component accounted for 5.92% of total variance, had an excellent overall model fit, F(3, 45) = 17.21, p < .0001, and was significantly associated with the contrast, Z = 5.67, p < .001. The loci of greatest activation were in left parietal lobe/postcentral gyrus (BA 3) and bilateral insula (BA 13). In addition, significant control > ideal-match activation was found in right superior frontal gyrus (BA 10) and bilateral putamen. The third component accounted for 5.11% of total variance, had a good overall model fit, F(3, 45) = 5.65, p < .01, and was significantly associated with scores on the promotion success scale, Z = 2.77, p < .01. The loci of greatest activation were in left middle frontal gyrus (BA 9 and 6), left superior frontal gyrus (BA 6), and right occipital gyrus (BA 19). In addition, significant deactivation was found in right middle frontal gyrus (BA 9). Figure 2 presents visual summaries of each of the three components for the ideal match TICA analysis.

Figure 2.

Spatial maps for components of interest in the ideal match versus control contrast analysis from TICA. The three components shown were selected via a priori criteria and are ordered by decreasing percent of total variance accounted for. Each map depicts axial views at z = 0, 10, 20, 30, 40, and 50 mm and is thresholded at Z ±2.30. Red/yellow signifies areas of significant activation; blue signifies areas of significant deactivation. Component A: 6.91% of variance, associated with both the contrast and promotion success scores. Component B: 5.92% of variance, associated with the contrast only. Component C: 5.11% of variance, associated with promotion success scores only.

2.2.2 Components associated with the ideal mismatch versus yoked-control contrast

We found three components of interest related to this contrast. The first component accounted for 9.48% of total variance, had an excellent overall model fit, F(3, 45) = 62.33, p < .0001, and was significantly associated with both the contrast per se and with scores on the promotion success scale, Z = 8.50 (p < .001) and 2.06 (p < .02), respectively. The loci of greatest activation were in bilateral occipital lobe (BA 18), left precentral gyrus (BA 6), and left parietal lobe (BA 40). In addition, significant deactivation was found in right parietal lobe (BA 3). The second component accounted for 8.30% of total variance, had an excellent overall model fit, F(3, 45) = 44.18, p < .0001, and was significantly associated with both the contrast per se and with scores on the promotion success scale, Z = 7.48 (p < .001) and 3.77 (p < .001), respectively. The loci of greatest activation were in bilateral temporal lobe (BA 22), bilateral occipital lobe (BA 18), bilateral superior frontal gyrus (BA 8), left middle frontal gyrus (BA 9), and right precentral gyrus (BA 4). No significant deactivation was observed. The third component accounted for 6.29% of total variance, had an excellent overall model fit, F(3, 45) = 14.74, p < .0001, and was significantly associated with both the contrast per se and with scores on the promotion success scale, Z = 5.41 (p < .001) and 2.08 (p < .02), respectively. The loci of greatest activation were in bilateral insula (BA 13). In addition, significant deactivation was found in bilateral precuneus (BA 7). Figure 3 presents visual summaries of each of the three components for the ideal mismatch TICA analysis.

Figure 3.

Spatial maps for components of interest in the ideal mismatch versus control contrast analysis from TICA. The three components shown were selected via a priori criteria and are ordered by decreasing percent of total variance accounted for. Each map depicts axial views at z = 0, 10, 20, 30, 40, and 50 mm and is thresholded at Z ±2.30. Red/yellow signifies areas of significant activation; blue signifies areas of significant deactivation. Component A: 9.48% of variance, associated with both the contrast and promotion success scores. Component B: 8.30% of variance, associated with both the contrast and promotion success scores. Component C: 6.29% of variance, associated with both the contrast and promotion success scores.

2.2.3 Components associated with the ought match versus yoked-control contrast

We found two components of interest related to this contrast. The first component accounted for 12.40% of total variance, had an excellent overall model fit, F(3, 45) = 17.83, p < .0001, and was significantly associated with the contrast, Z = 5.76, p < .001. The loci of greatest activation were in left precentral gyrus (BA 3 and 4) and bilateral medial frontal gyrus (BA 6), with no regions showing significant deactivation. The second component accounted for 6.01% of total variance, had a good overall model fit, F(3, 45) = 4.98, p < .01, and was significantly associated with the contrast, Z = 3.82, p < .001. The loci of greatest activation were in right superior frontal gyrus (BA 8), bilateral insula (BA 13), and right frontal gyrus (BA 6). In addition, significant deactivation was found in right occipital gyrus (BA 18). Figure 4 presents visual summaries of each of the two components for the ought match TICA analysis.

Figure 4.

Spatial maps for components of interest in the ought match versus control contrast analysis from TICA. The two components shown were selected via a priori criteria and are ordered by decreasing percent of total variance accounted for. Each map depicts axial views at z = 0, 10, 20, 30, 40, and 50 mm and is thresholded at Z ±2.30. Red/yellow signifies areas of significant activation; blue signifies areas of significant deactivation. Component A: 12.40% of variance, associated with the contrast only. Component B: 6.01% of variance, associated with the contrast only.

2.2.4 Components associated with the ought mismatch versus yoked-control contrast

Using our selection criteria, we found two components of interest related to this contrast. The first component accounted for 9.87% of total variance, had an excellent overall model fit, F(3, 45) = 32.18, p < .0001, and was significantly associated with the contrast, Z = 7.11, p < .001. The loci of greatest activation were in bilateral occipital lobe (BA 18). There was also significant deactivation in left frontal gyrus (BA 6). The second component accounted for 6.75% of total variance, had a good overall model fit, F(3, 45) = 4.46, p < .01, and was significantly associated with the contrast, Z = 3.34, p < .001. The loci of greatest activation were in bilateral insula (BA 13) and left temporal gyrus (BA 39). In addition, significant deactivation was found in left middle frontal gyrus (BA 10), bilateral occipital lobe (BA 18), and right superior temporal gyrus (BA 42). Figure 5 presents visual summaries of each of the two components for the ought mismatch TICA analysis.

Figure 5.

Spatial maps for components of interest in the ought mismatch versus control contrast analysis from TICA. The two components shown were selected via a priori criteria and are ordered by decreasing percent of total variance accounted for. Each map depicts axial views at z = 0, 10, 20, 30, 40, and 50 mm and is thresholded at Z ±2.30. Red/yellow signifies areas of significant activation; blue signifies areas of significant deactivation. Component A: 9.87% of variance, associated with the contrast only. Component B: 6.75% of variance, associated with the contrast only.