Neural Responses to Maternal Criticism in Healthy Youth
Participants were 32 healthy adolescents [22 women, aged 9–17 years (M = 14.34, s.d. = 2.04)]. They were recruited from community advertisements, pediatric offices and existing research projects. Exclusion criteria for the study included: (i) current or lifetime DSM-IV (1994) diagnosis for any Axis 1 disorder, (ii) the existence of a major systemic medical illness, (iii) a history of serious head injury and (iv) presence of metal objects in the body. One participant did not complete this task due to an equipment problem and three participants were excluded due to excessive head movement [over 30% of scans with greater than ±5 mm and ±5° movement from a reference image and ±1 mm and ±1° incremental (scan-to-scan) movement]. Twenty-eight typically developing youth [20 women, aged 9–17 years (M = 14.57, s.d. = 1.95)] were thus included for our final analysis.
Participants provided informed consent using a form approved by the University of Pittsburgh Institutional Review Board. Participants completed two laboratory visits. During the first visit, participants completed a structured diagnostic interview, and the mothers recorded audio clips to be used during the fMRI assessment. The fMRI assessment was completed during their second visit.
Structured Diagnostic Interviews. On their first visit to the laboratory, each youth and his or her parent(s) were interviewed to determine the youth's mental health history using the Schedule for Affective Disorders and Schizophrenia in School-Age Children—Present and Lifetime version (Kaufman et al., 1997). Parents and youth were interviewed separately, with clinicians integrating data from both informants to arrive at a final diagnosis. All interviews were carried out by trained Bachelor of Arts (BA) level and Master of Arts (MA)-level clinicians. Fifteen percent of interviews were double coded and there were no diagnostic disagreements (kappa = 1.0).
fMRI Assessment and Debriefing. Participants underwent an fMRI scan. They were asked to lie still as possible during the structural imaging acquisition and then to listen to their mother's audio clips or rest during the functional imaging acquisition. To be able to relate brain activity to subjective reactivity to the comments, after the fMRI assessment, participants were asked to respond to two questions (post-scan emotion ratings): 'How negative was the comment?' and 'How upset did it make you feel?' The rating scale ranged from 1 (not at all) to 10 (very). Participants were carefully debriefed following completion of the scan.
Self-report Measure. Parental warmth was assessed by a shortened version of the Child Report of Parent Behavior Inventory (CRPBI) (Schludermann and Schludermann, 1970). Participants responded to each question about their parents' behavior on a three-point scale, ranging from 1 (not like your mother) to 3 (a lot like your mother).
During the fMRI scan, participants were asked to hear their own mother's comments about them. There were two audio clips for critical, praising and neutral comments. Each audio clip lasted for 30 s. To present audio clips as clearly as possible during scanning, the comments were delivered via MRI compatible headphones. We first tested whether participants could hear comments clearly using a sample audio clip recorded by the mother prior to scanning. These audio clips were recorded by the participant's own mother during the first visit. We followed similar procedures used in previous studies (Hooley et al., 2005, 2009) for obtaining audio clips. Each mother was asked to produce two 30 s audio clips describing things that bother her about her child [critical statements beginning with '(Child's name), one thing that bothers me about you is …', e.g. not doing chores or attitudes toward family member(s)], two 30 s audio clips describing things that she especially likes about her child [praising statements beginning with '(Child's name), one thing I really like about you is …', i.e. sense of humor, being a nice person, willingness to help out and academic and extracurricular achievements] and two 30 s neutral clips (neutral statements: something your child won't find interesting, i.e. grocery shopping, parent work or chores, and weather). Examples of mother's critical and neutral comments are presented in Table 1. To create these clips, each mother was instructed to formulate their critical remarks based on something they have shared with their child on more than one occasion, so that youth would not be exposed to new and potentially disturbing information in the scanner. Praising comments were included to balance critical remarks and to mitigate potential negative effects of critical comments during the scanning.
There was one block each for critical, praising and neutral conditions. Each block (run) consisted of two 30.06 s comment presentations (30 s audio clip with 0.06 s additional duration to match with our TR 1.67 s) and three 30.06 s rest periods. Each began with a 30.06 s rest period, followed by 30.06 s of one's own mother's comment presentation, the second rest period, another comment presentation and then the last rest period. Participants were scanned both when they heard their own mother's comment and when they were at rest. To minimize possible emotional carryover after listening to criticism or praise from parents, the neutral condition block was presented first and the order of two other condition blocks were counterbalanced across participants.
Imaging Acquisition. Images were acquired on a 3 T Trio scanner (Siemens, Erlangen, Germany). Thirty-two 3.2 mm slices were acquired parallel to the AC–PC line using a posterior-to-anterior echo planar pulse sequence (T2*-weighted image depicting BOLD signal; Repetition Time (TR) = 1670 ms, Echo Time (TE) = 29 ms, Field of View (FOV) = 205 mm, flip angle = 75). Each image was acquired in 1.67 s, allowing 18 scans per 30.06 s trial consisting of either a 30.06 s comment presentation or rest period. There were three blocks (runs). Each block (run) lasted for 150.3 s (2.505 min). Ninety images (150.3 s/TR = 1.67 s) were collected in each block (run), consequently total 270 images were acquired. High-resolution T1-weighted MPRAGE images (1 mm, axial) were also collected for use in cross-registration.
fMRI Data Preprocessing. fMRI analyses were conducted using locally developed NeuroImaging Software (NIS) (Fissell et al., 2003) and Analysis of Functional Neuroimaging (AFNI) software (Cox, 1996). Functional imaging data were corrected for motion using 3dVolReg implemented in AFNI using the first image as a reference. Quadratic trends within runs were removed and outliers over 1.5 interquartile range from the 25th or 75th percentiles were Winsorized using niscorrect from NIS to remove non-physiological spikes. Data were temporally smoothed using a four-point Gaussian filter and converted to %-change based on the median of all imaging data. Data were co-registered to the Colin-27 Montreal Neurological Institute template using the Automated Image Registration package's (Woods et al., 1993) 32-parameter non-linear automated warping algorithm and spatially smoothed using a 6 mm full width at half maximum filter.
Manipulation Check: Post-scan Ratings. Paired-samples t-tests were conducted to compare post-scan ratings following maternal critical vs neutral comments to ascertain the affective value of the comments.
Whole-brain Analyses: Neural Responses to Maternal Criticism Compared With Neutral Comments. To examine temporal dynamics of neural responses to maternal critical and neutral comments over time, a random-effects whole-brain voxelwise analysis of variance (ANOVA) was conducted with participant as a random factor, and valence (criticism vs neutral) and time (18 scans of 0–30.06 s in each trial of commentary) as fixed factors. This model-free analysis was employed to account for empirical variation in the shape of the hemodynamic response (e.g. sustained activity or early deactivation) rather than relying on hemodynamic responses to have a canonical shape. To control for temporal autocorrelation, brain regions identified from the whole-brain analysis were further subjected to mixed-effects analyses of signal change using valence (criticism vs neutral remarks) and time as repeated measures and subject as a random factor, assuming an AR1 covariance structure using restricted maximum-likelihood estimation (REML).
To further understand temporal dynamics of brain activity, we found specific temporal regions (e.g. sustained activity at 21–30 s after the onset of criticism) showing significant differences in time courses between critical and neutral statements by comparing time courses between two conditions at each time point (scan). Guthrie and Buchwald's (1991) method was used to control type I error when point-by-point tests in entire time courses were performed to detect significantly different temporal regions at P < 0.05. Thus, temporal regions with significant differences between two conditions represented continuous series of time points that reliably differed in time courses.
The valence × time interaction effect map was thresholded at an uncorrected P < 0.0001. To control type 1 error at P < 0.05 across the whole brain for each family of tests (i.e. <5% chance that even one voxel was identified in error), voxelwise tests at a given statistical threshold (P < 0.0001) were subjected to empirically determine contiguity thresholds based on the spatial autocorrelation of statistical maps using AFNI's AlphaSim program. We used a conservative voxelwise threshold (e.g. P < 0.0001) that requires a small number of voxels contiguity to find small regions of our hypothesized brain areas such as amygdala. Thus, both the uncorrected P value and contiguity threshold necessary to achieve a corrected brain-wise P < 0.05 were reported with each test described below.
Region of Interest (ROI) Analysis: Brain Activity in Regions Reacting to Criticism During the Rest Period. To further understand whether neural responses to maternal comments last during the rest period, time courses in brain regions (nine functional ROIs shown in Figure 1) identified from the valence × time interaction effect and proposed in our model were extracted during the rest period. Mixed-effects analyses were used to test significant differences in time courses with condition (rest periods after criticism vs rest periods after neutral comments) and time as repeated measures and subject as a random factor, assuming an AR1 covariance structure using REML.
(Enlarge Image)
Figure 1.
Neural responses to maternal criticism compared with neutral remarks: time courses in brain regions within affective, cognitive control and social cognitive networks were plotted when hearing own mother's critical and neutral statements (audio clip: 0–30.06 s) and the rest period (30.07–60.12 s). (a) Brain regions in affective networks showed increased and sustained activity when hearing own mother's criticism (bilateral lentiform nucleus and posterior insula) and during the rest period (right lentiform nucleus). (b) Brain regions in cognitive control networks showed decreased activity in response to maternal criticism (bilateral DLPFC and cACC). (c) Brain regions in social cognitive networks showed decreased neural response to maternal criticism (bilateral TPJ/IPL and PCC/precuneus). Significant differences in time courses between critical and neutral statements are highlighted below the x axis (pink: P < 0.05). Note. Average activity from the entire ROI was used to plot each time series.
Association With Perceived Negativity, Parental Warmth and Age. We tested associations of neural responses to criticism with affective-social (i.e. perceived negativity and parental warmth) and developmental (i.e. age) factors by correlating post-scan ratings of perceived negativity, parental warmth and age and averaged brain activity across the temporal regions that displayed significant condition (criticism vs neutral) differences and were highlighted below the x axis of each brain region in Figure 1.
Relationships Between Brain Networks Proposed in Our Model. We proposed bi-directional relationships between brain networks involved in the processing of maternal criticism in our model. These relationships were examined using the psychophysiological interaction analysis (PPI; Friston et al., 1997; O'Reilly et al., 2012) adapted for AFNI (http://afni.nimh.nih.gov/sscc/gangc/CD-CorrAna.html). PPI analysis allowed us to examine significant changes in functional connectivity between brain regions during maternal criticism compared with neutral comments. Our seed brain regions were nine functional ROIs (Figure 1) identified from our primary ANOVA (valence × time) analysis. The time series extracted from each seed brain region was multiplied by experimental conditions (i.e. criticism = 1, neutral condition = −1 and other conditions = 0) to create a PPI variable. For each subject, regression analyses implemented in AFNI's 3dDeconvolve were conducted by entering the physiological variable (time course of seed region), psychological variable (critical vs neutral comments) and their interaction (PPI) variable as regressors. To model only critical and neural comments, time points associated with other conditions (i.e. praise and rest periods) were further censored out. The resulting correlation coefficient (r) scores were converted to z scores through a Fisher's r to z transformation to perform group analyses using one-sample t-tests. To control for multiple comparisons, both the uncorrected P value and contiguity thresholds based on the spatialautocorrelation of our functional ROI masks using AFNI's AlphaSim were necessary to achieve a corrected P < 0.05. Thus, using an uncorrected threshold of P < 0.005, to achieve a corrected type I error of α=0.05 with a small volume correction, cluster sizes (3–31 voxels contiguity) were required.
Methods
Participants
Participants were 32 healthy adolescents [22 women, aged 9–17 years (M = 14.34, s.d. = 2.04)]. They were recruited from community advertisements, pediatric offices and existing research projects. Exclusion criteria for the study included: (i) current or lifetime DSM-IV (1994) diagnosis for any Axis 1 disorder, (ii) the existence of a major systemic medical illness, (iii) a history of serious head injury and (iv) presence of metal objects in the body. One participant did not complete this task due to an equipment problem and three participants were excluded due to excessive head movement [over 30% of scans with greater than ±5 mm and ±5° movement from a reference image and ±1 mm and ±1° incremental (scan-to-scan) movement]. Twenty-eight typically developing youth [20 women, aged 9–17 years (M = 14.57, s.d. = 1.95)] were thus included for our final analysis.
Procedure
Participants provided informed consent using a form approved by the University of Pittsburgh Institutional Review Board. Participants completed two laboratory visits. During the first visit, participants completed a structured diagnostic interview, and the mothers recorded audio clips to be used during the fMRI assessment. The fMRI assessment was completed during their second visit.
Structured Diagnostic Interviews. On their first visit to the laboratory, each youth and his or her parent(s) were interviewed to determine the youth's mental health history using the Schedule for Affective Disorders and Schizophrenia in School-Age Children—Present and Lifetime version (Kaufman et al., 1997). Parents and youth were interviewed separately, with clinicians integrating data from both informants to arrive at a final diagnosis. All interviews were carried out by trained Bachelor of Arts (BA) level and Master of Arts (MA)-level clinicians. Fifteen percent of interviews were double coded and there were no diagnostic disagreements (kappa = 1.0).
fMRI Assessment and Debriefing. Participants underwent an fMRI scan. They were asked to lie still as possible during the structural imaging acquisition and then to listen to their mother's audio clips or rest during the functional imaging acquisition. To be able to relate brain activity to subjective reactivity to the comments, after the fMRI assessment, participants were asked to respond to two questions (post-scan emotion ratings): 'How negative was the comment?' and 'How upset did it make you feel?' The rating scale ranged from 1 (not at all) to 10 (very). Participants were carefully debriefed following completion of the scan.
Self-report Measure. Parental warmth was assessed by a shortened version of the Child Report of Parent Behavior Inventory (CRPBI) (Schludermann and Schludermann, 1970). Participants responded to each question about their parents' behavior on a three-point scale, ranging from 1 (not like your mother) to 3 (a lot like your mother).
Stimuli and Experimental Paradigms
During the fMRI scan, participants were asked to hear their own mother's comments about them. There were two audio clips for critical, praising and neutral comments. Each audio clip lasted for 30 s. To present audio clips as clearly as possible during scanning, the comments were delivered via MRI compatible headphones. We first tested whether participants could hear comments clearly using a sample audio clip recorded by the mother prior to scanning. These audio clips were recorded by the participant's own mother during the first visit. We followed similar procedures used in previous studies (Hooley et al., 2005, 2009) for obtaining audio clips. Each mother was asked to produce two 30 s audio clips describing things that bother her about her child [critical statements beginning with '(Child's name), one thing that bothers me about you is …', e.g. not doing chores or attitudes toward family member(s)], two 30 s audio clips describing things that she especially likes about her child [praising statements beginning with '(Child's name), one thing I really like about you is …', i.e. sense of humor, being a nice person, willingness to help out and academic and extracurricular achievements] and two 30 s neutral clips (neutral statements: something your child won't find interesting, i.e. grocery shopping, parent work or chores, and weather). Examples of mother's critical and neutral comments are presented in Table 1. To create these clips, each mother was instructed to formulate their critical remarks based on something they have shared with their child on more than one occasion, so that youth would not be exposed to new and potentially disturbing information in the scanner. Praising comments were included to balance critical remarks and to mitigate potential negative effects of critical comments during the scanning.
There was one block each for critical, praising and neutral conditions. Each block (run) consisted of two 30.06 s comment presentations (30 s audio clip with 0.06 s additional duration to match with our TR 1.67 s) and three 30.06 s rest periods. Each began with a 30.06 s rest period, followed by 30.06 s of one's own mother's comment presentation, the second rest period, another comment presentation and then the last rest period. Participants were scanned both when they heard their own mother's comment and when they were at rest. To minimize possible emotional carryover after listening to criticism or praise from parents, the neutral condition block was presented first and the order of two other condition blocks were counterbalanced across participants.
Imaging Acquisition and Preprocessing
Imaging Acquisition. Images were acquired on a 3 T Trio scanner (Siemens, Erlangen, Germany). Thirty-two 3.2 mm slices were acquired parallel to the AC–PC line using a posterior-to-anterior echo planar pulse sequence (T2*-weighted image depicting BOLD signal; Repetition Time (TR) = 1670 ms, Echo Time (TE) = 29 ms, Field of View (FOV) = 205 mm, flip angle = 75). Each image was acquired in 1.67 s, allowing 18 scans per 30.06 s trial consisting of either a 30.06 s comment presentation or rest period. There were three blocks (runs). Each block (run) lasted for 150.3 s (2.505 min). Ninety images (150.3 s/TR = 1.67 s) were collected in each block (run), consequently total 270 images were acquired. High-resolution T1-weighted MPRAGE images (1 mm, axial) were also collected for use in cross-registration.
fMRI Data Preprocessing. fMRI analyses were conducted using locally developed NeuroImaging Software (NIS) (Fissell et al., 2003) and Analysis of Functional Neuroimaging (AFNI) software (Cox, 1996). Functional imaging data were corrected for motion using 3dVolReg implemented in AFNI using the first image as a reference. Quadratic trends within runs were removed and outliers over 1.5 interquartile range from the 25th or 75th percentiles were Winsorized using niscorrect from NIS to remove non-physiological spikes. Data were temporally smoothed using a four-point Gaussian filter and converted to %-change based on the median of all imaging data. Data were co-registered to the Colin-27 Montreal Neurological Institute template using the Automated Image Registration package's (Woods et al., 1993) 32-parameter non-linear automated warping algorithm and spatially smoothed using a 6 mm full width at half maximum filter.
Statistical Analyses
Manipulation Check: Post-scan Ratings. Paired-samples t-tests were conducted to compare post-scan ratings following maternal critical vs neutral comments to ascertain the affective value of the comments.
Whole-brain Analyses: Neural Responses to Maternal Criticism Compared With Neutral Comments. To examine temporal dynamics of neural responses to maternal critical and neutral comments over time, a random-effects whole-brain voxelwise analysis of variance (ANOVA) was conducted with participant as a random factor, and valence (criticism vs neutral) and time (18 scans of 0–30.06 s in each trial of commentary) as fixed factors. This model-free analysis was employed to account for empirical variation in the shape of the hemodynamic response (e.g. sustained activity or early deactivation) rather than relying on hemodynamic responses to have a canonical shape. To control for temporal autocorrelation, brain regions identified from the whole-brain analysis were further subjected to mixed-effects analyses of signal change using valence (criticism vs neutral remarks) and time as repeated measures and subject as a random factor, assuming an AR1 covariance structure using restricted maximum-likelihood estimation (REML).
To further understand temporal dynamics of brain activity, we found specific temporal regions (e.g. sustained activity at 21–30 s after the onset of criticism) showing significant differences in time courses between critical and neutral statements by comparing time courses between two conditions at each time point (scan). Guthrie and Buchwald's (1991) method was used to control type I error when point-by-point tests in entire time courses were performed to detect significantly different temporal regions at P < 0.05. Thus, temporal regions with significant differences between two conditions represented continuous series of time points that reliably differed in time courses.
The valence × time interaction effect map was thresholded at an uncorrected P < 0.0001. To control type 1 error at P < 0.05 across the whole brain for each family of tests (i.e. <5% chance that even one voxel was identified in error), voxelwise tests at a given statistical threshold (P < 0.0001) were subjected to empirically determine contiguity thresholds based on the spatial autocorrelation of statistical maps using AFNI's AlphaSim program. We used a conservative voxelwise threshold (e.g. P < 0.0001) that requires a small number of voxels contiguity to find small regions of our hypothesized brain areas such as amygdala. Thus, both the uncorrected P value and contiguity threshold necessary to achieve a corrected brain-wise P < 0.05 were reported with each test described below.
Region of Interest (ROI) Analysis: Brain Activity in Regions Reacting to Criticism During the Rest Period. To further understand whether neural responses to maternal comments last during the rest period, time courses in brain regions (nine functional ROIs shown in Figure 1) identified from the valence × time interaction effect and proposed in our model were extracted during the rest period. Mixed-effects analyses were used to test significant differences in time courses with condition (rest periods after criticism vs rest periods after neutral comments) and time as repeated measures and subject as a random factor, assuming an AR1 covariance structure using REML.
(Enlarge Image)
Figure 1.
Neural responses to maternal criticism compared with neutral remarks: time courses in brain regions within affective, cognitive control and social cognitive networks were plotted when hearing own mother's critical and neutral statements (audio clip: 0–30.06 s) and the rest period (30.07–60.12 s). (a) Brain regions in affective networks showed increased and sustained activity when hearing own mother's criticism (bilateral lentiform nucleus and posterior insula) and during the rest period (right lentiform nucleus). (b) Brain regions in cognitive control networks showed decreased activity in response to maternal criticism (bilateral DLPFC and cACC). (c) Brain regions in social cognitive networks showed decreased neural response to maternal criticism (bilateral TPJ/IPL and PCC/precuneus). Significant differences in time courses between critical and neutral statements are highlighted below the x axis (pink: P < 0.05). Note. Average activity from the entire ROI was used to plot each time series.
Association With Perceived Negativity, Parental Warmth and Age. We tested associations of neural responses to criticism with affective-social (i.e. perceived negativity and parental warmth) and developmental (i.e. age) factors by correlating post-scan ratings of perceived negativity, parental warmth and age and averaged brain activity across the temporal regions that displayed significant condition (criticism vs neutral) differences and were highlighted below the x axis of each brain region in Figure 1.
Relationships Between Brain Networks Proposed in Our Model. We proposed bi-directional relationships between brain networks involved in the processing of maternal criticism in our model. These relationships were examined using the psychophysiological interaction analysis (PPI; Friston et al., 1997; O'Reilly et al., 2012) adapted for AFNI (http://afni.nimh.nih.gov/sscc/gangc/CD-CorrAna.html). PPI analysis allowed us to examine significant changes in functional connectivity between brain regions during maternal criticism compared with neutral comments. Our seed brain regions were nine functional ROIs (Figure 1) identified from our primary ANOVA (valence × time) analysis. The time series extracted from each seed brain region was multiplied by experimental conditions (i.e. criticism = 1, neutral condition = −1 and other conditions = 0) to create a PPI variable. For each subject, regression analyses implemented in AFNI's 3dDeconvolve were conducted by entering the physiological variable (time course of seed region), psychological variable (critical vs neutral comments) and their interaction (PPI) variable as regressors. To model only critical and neural comments, time points associated with other conditions (i.e. praise and rest periods) were further censored out. The resulting correlation coefficient (r) scores were converted to z scores through a Fisher's r to z transformation to perform group analyses using one-sample t-tests. To control for multiple comparisons, both the uncorrected P value and contiguity thresholds based on the spatialautocorrelation of our functional ROI masks using AFNI's AlphaSim were necessary to achieve a corrected P < 0.05. Thus, using an uncorrected threshold of P < 0.005, to achieve a corrected type I error of α=0.05 with a small volume correction, cluster sizes (3–31 voxels contiguity) were required.