Abstract
Introduction
Previous research has shown that performance on cognitive tasks administered in the scanner can be altered by the scanner environment. There are no previous studies that have investigated the impact of scanner noise using a well-validated measure of affective change. The goal of this study was to determine whether performance on an affective attentional task or emotional response to the task would change in the presence of distracting acoustic noise, such as that encountered in an MRI environment.
Method
Thirty-four young adults with no self-reported history of neurologic disorder or mental illness completed three blocks of the affective Posner task outside of the scanner. The task was meant to induce frustration through monetary contingencies and rigged feedback.
Participants completed a self-assessment manikin at the end of each block to rate their mood, arousal level, and sense of dominance. During the task, half of the participants heard noise (recorded from a 4T MRI system), and half heard no noise.
Results
The affective Posner task led to significant reductions in mood and increases in arousal in healthy participants. The presence of scanner noise did not impact task performance; however, individuals in the noise group did report significantly poorer mood throughout the task.
Conclusions
The results of the present study suggest that the acoustic qualities of MRI enhance frustration effects on an affective attentional task and that scanner noise may influence mood during similar fMRI tasks.
Keywords: Attention, affect, imaging, fMRI, noise
Functional magnetic resonance imaging (fMRI) has become an increasingly common tool for studies of cognitive processes. Factors inherent to the process of fMRI acquisition, such as loud noise, supine body position, and being in an enclosed environment have been shown to impact affective and neural processes outside of the scanner and, therefore, may affect fMRI findings. For example, laying flat on one's back while undergoing electroencephalography (EEG) reduces neural responses to anger cues (Harmon-Jones & Peterson, 2009). Additionally, the enclosed scanner environment leads to unpleasant feelings including anxiety and fear (Melendez & McCrank, 1993), and loud acoustic noise increases negative emotions such as anger, fear, and depression (Breier et al., 1987). The scanner environment has also been linked to higher self-reported anxiety and increased cortisol levels in individuals exposed to it for the first time (Tessner, Walker, Hockman, & Hamann, 2006). Taken together, these findings suggest that the fMRI image acquisition process may systematically bias certain behavioral and neuroendocrine research measures related to negative affect. Furthermore, there is a large literature documenting the deleterious effect of acoustic noise on performance accuracy (for meta-analytic review see Szalma & Hancock, 2010). Given the potential impact of acoustic noise on cognitive processes that are often of interest in research studies, the present study focused on the impact of scanner noise on affect and attentional performance.
The acoustic noise that accompanies fMRI image acquisition has been shown to have direct and indirect effects on imaging results. The direct effects involve increased blood oxygen level-dependent (BOLD) responses in the auditory cortex that could confound auditory activation measurements. The indirect effects are thought to be predominantly attention related, with increased activity in attentional areas and decreased cortical activity in task related areas (see Moelker & Pattynama, 2003 for review). Increased activation in attentional areas may arise from these brain areas compensating due to the distraction created by the noise. In a study investigating the effects of scanner noise on brain activation during a visual verbal working memory (N-back) task, healthy participants underwent a 4T fMRI and completed the task in both “quiet” and “loud” conditions (Tomasi, Caparelli, Chang, & Ernst 2005). Although there was no difference between the two noise conditions in performance accuracy or reaction time, increased acoustic noise was associated with increased activation in attention networks (Tomasi et al., 2005). The authors suggested that in the context of preserved accuracy and reaction time, this increase in activation indicated compensation for auditory interference on cognitive processing.
A related line of evidence links acoustic noise to increased cognitive control and improved behavioral performance. In a series of three experiments performed outside of the scanner that investigated cognitive control processes (i.e., interference inhibition, task switching, and response inhibition) the presence of scanner noise positively impacted cognitive control (Hommel, Fischer, Colzato, Wildenberg, & Cellini, 2012). The improved cognitive control in the presence of scanner noise alone (i.e., no other aspects of the scanner environment were present) is, perhaps, counterintuitive, but has substantial implications for behavioral data collected in the scanner.
Mazzard and colleagues (2002) investigated the impact of scanner noise on visual-imagery task performance and neural activation. The presence of fMRI noise led to poorer performance on the visual mental imagery task and to increased activation in the anterior cingulate cortex and Wernike's area (Mazzard et al., 2002). These findings are inconsistent with findings of preserved performance in the presence of acoustic noise noted above, suggesting that the increased attentional load may be detrimental to performance on tasks that are more dependent on cognitive resources outside of attention and cognitive control. Additionally, increased neural activation was present in areas that involve performance monitoring and processing of verbal stimuli, implying that increased load impaired attentional and auditory processes in task relevant areas (Mazzard., 2002).
Our study is predicated on the assumption that a modified affective Posner task would have the same effects on affective responses in healthy participants as have been shown in psychiatric populations. We investigated whether performance on a task designed to induce frustration would change in the presence of distracting acoustic noise, such as that encountered in an MRI environment. We hypothesized that participants listening to noise recorded from an MRI scanner would perform more poorly on the task than those who were not listening to distracting acoustic noise. It was also hypothesized that the presence of scanner noise would have a negative impact on affective responses. Similar paradigms may be useful in the future to study interactions between affective and cognitive neural networks in healthy subjects and psychiatric populations.
Method
Participants
Thirty-four college students, ages 18-29, participated in this study (71% female; 79% white; 97% right handed) in exchange for course credit. All participants reported healthy hearing and corrected to healthy vision. Participants were without self-reported history of neurologic, psychiatric, or substance abuse disorders or any other serious medical condition that could affect cognition. The study was approved by the University of Cincinnati Institutional Review Board, and written informed consent was obtained from all participants before testing.
Experimental Design
Participants were randomly assigned to the noise (n = 17) or no-noise (n = 17) groups. The noise group heard sounds through headphones that were recorded from inside the bore of a 4T MRI system running an echo planar data acquisition sequence that is commonly used during fMRI protocols; the no-noise group wore headphones but heard no noise. There were no group differences in demographics (p < .05 for all comparisons; see Table 1). All testing took place in the Neuropsychology and Social Cognition Laboratory in the psychology department at the University of Cincinnati. Sitting in an upright position, participants completed three 50-trial blocks of a modified Affective Posner Task (Rich et al., 2005), which was designed to induce frustration across blocks. Participants completed the task sitting 12 inches from a 10×16 PC display. The task was programmed in the E-Prime application suite, and the program recorded reaction time and accuracy.
Table 1.
Participant demographics by condition
| No Noise (n = 17) | Noise (n = 17) | |
|---|---|---|
| Age M(SD) | 21.29(2.80) | 20.82(2.04) |
| Gender (% male) | 29 | 29 |
| Ethnicity (% white) | 88 | 70.5 |
| Handedness (% right) | 100 | 94 |
Measures
The Posner task (Posner, 1980) is a well-established paradigm to investigate spatial attention. Participants are asked to respond to a spatial target that is preceded by a spatial cue. The cue directs the individual's attention toward the target (valid cue), away from the target (invalid cue), or towards a neutral location (neutral cue). It is well documented that individuals take longer to respond when the target is preceded by an invalid cue compared to valid or neutral conditions. The affective Posner Task (Rich et al., 2005) is a modification of this paradigm that induces frustration by providing negative feedback and incorporating monetary contingencies.
Participants completed three, 50-trial blocks of a standard Posner Task. Participants were told to place their index and middle finger on the “B” and “N” keys on the keyboard, respectively. They saw black outlines of three squares and were told to be prepared to respond. Next, the cue (blue square) flashed briefly in the left, right, or center square, and they were told not to respond to the flash. The cue either appeared in the same square as the target (valid cue; n = 20), in the opposite square (invalid cue; n = 20); or in the middle square (neutral cue; n = 10). After the cue, the target (black circle) appeared in the left or right square. Participants were told to respond as quickly and as accurately as possible by pressing the right (“N”) key when the black circle appeared in the right square or pressing the left (“B) key when the target appeared in the left square. After their response, the screen went blank until the beginning of the next trial when the outlines of squares reappeared (see Figure 1). The final units of analyses were mean reaction time for all responses and total errors for each block. The three blocks varied in the type of feedback given and the introduction of monetary contingencies. In each case, the instructions appeared on the screen at the beginning of the block and were also read aloud by the researcher.
Figure 1.
Affective Posner Task. The second square illustrates the cue (blue square) and the third square illustrates the target (black circle), to which the participant respons.
Block 1
Participants received the following feedback regarding their performance: “Correct” for correct responses, “Incorrect” for incorrect responses, and “Too Slow” if the participant took longer than 1000 ms to respond. At the conclusion of Block 1 participants were shown a prompt on the computer screen: “Congratulations, you're fast! Advance to the next level.”
Block 2
During this block, a monetary contingency was introduced, and participants received the following feedback after each trial: “Correct, Earn 10 Cents” for correct responses, “Incorrect, Lose 10 Cents” for incorrect responses, and “Too Slow, Lose 10 Cents” if the participant took longer than 1000 ms to respond. At the conclusion of Block 2 participants were shown a prompt on the computer screen: “Sorry, you're too slow. Please repeat this level,” which was meant to induce frustration.
Block 3
The final block of the task also involved monetary contingencies; however, rigged feedback was provided with the goal of inducing frustration. During this block, a reaction time threshold was computed for each individual participant. The threshold was defined as 90% of the mean reaction time (RT) from all previously completed correct trials from Blocks 2 and 3 (threshold = .90*correct mean RT). In order to make the task increasingly difficult and induce frustration, the new mean RT for each trial was calculated using only correct responses with sub-threshold RT. The following feedback followed each trial: “Correct, Earn 10 Cents” for correct responses with RT less than the threshold value, “Too slow, lose ten cents” for correct responses with RT greater than the threshold value and “Incorrect, lose ten cents” for incorrect responses.
At the end of each block, noise was paused for the participants who were assigned to the noise condition. All participants were presented with a self-assessment manikin that depicted three aspects of emotion: mood, level of arousal, and dominance (see Figure 2). The dominance measure is associated with one having maximum control in a situation (Bradley & Lang, 1994). Although it is a rather vague construct and the authors recognize that in previous research it accounts for the least amount of variance in affect (Bradley & Lang, 1994), it was included in this study in order to keep our measure of affect consistent with the original self-assessment manikin. For each of three aspects of emotion the participants rated their current feelings along a continuum by moving an arrow on the screen, which indicated a value on the corresponding line. The center point on each continuum was defined as having a value of zero, and self-ratings were computed as the distance in pixels between that zero point and the arrow placement, with a range of −320 to 320. Poorer mood (lower pixel values), higher arousal level (higher pixel values), and decreased degree of dominance (lower pixel values) were posited to represent increased frustration.
Figure 2.
Self-assessment manikin. Mood, arousal, and dominance are represented in descending order.
In order to make participants believe that they were being paid based on the accuracy of their performance, they were told by the researcher that they would earn one research credit for their participation and that they would have the opportunity to earn money based on their performance during blocks two and three of the task. Regardless of their performance, all participants received equal compensation. Upon completion of the tasks, all participants were fully debriefed.
Statistical Analyses
Although the task was developed to include 50 trials in each of the three blocks, the first 16 participants (n = 8 with noise; n = 8 with no noise) completed only 15 trials in the first block due to a programming error. In order to ensure that the Block 1 paradigm (i.e., 15 trials vs. 50 trials) did not impact the task performance and affective data, two statistical approaches were taken: 1) two by two univariate ANOVAs were run in order to make sure there was no effect of paradigm on performance or frustration as well as no interaction between paradigm and noise condition; 2) paradigm was included in the repeated measures ANOVAs as a covariate in order to confirm that significant effects remained when paradigm was removed from the model. There was no effect of paradigm on accuracy, reaction time, or affective data, and therefore the following analyses were run across Block 1 paradigms. All data were investigated for sphericity and normal distribution. On the one analysis that violated sphericity assumptions (the effect of task Block on performance accuracy), the significant findings remained unchanged after applying the Greenhouse-Geisser correction. A series of repeated measures ANOVAs was used to evaluate the effect of block on reaction time, accuracy, and self-reported frustration, with noise as a grouping variable. Given the small error rate, McNemar's test was used to compare the frequency of errors across blocks. The pairwise comparisons between the three blocks were carried out using the Bonferroni adjustment.
Results
Effects of Block and Noise on Task Performance
There was a significant effect of block on performance accuracy, F(2, 31) = 19.06, p < .05, ηp2 = .37. Participants committed significantly more errors on Block 3 (M = 1.85, SD = 1.89) than on either Block 1 (M = .41, SD = .78) or Block 2 (M = .26, SD = .67), p < .05 for all comparisons. The error rate on all three blocks was extremely small. McNemar's test revealed that significantly more participants (p < .05) made at least one error on Block 3 than on either Block 1 or Block 2. In terms of reaction time (ms), there was a significant effect of block, F(2, 31) = 95.10, p < .05, ηp2 = .77 (see Figure 3), with participants responding significantly more quickly when the rigged feedback and reaction time threshold were introduced on Block 3 (M = 328.31, SD = 9.23) than on either Block 1 (M = 416.61, SD = 14.87) or Block 2 (M = 400.35, SD = 15.28; p < .05 for all comparisons), which did not differ. There was not a significant main effect of noise on either accuracy or reaction time (p > .05 for both), nor were there significant interactions between noise condition and block (p > .05 for both).
Figure 3.
Reaction time across blocks in the noise and no-noise conditions. Standard error bars are represented.
Effects of Block and Noise on Affect
On the self-report manikins, there were significant effects of block on mood, F(2, 31) = 10.29, p < .05, ηp2 = .24 (see Figure 4.A) and arousal, F(2, 31) = 20.09, p < .05, ηp2 = .39 (see Figure 4.B). Lower mood and greater arousal were reported after Block 3 (rigged feedback) than after Blocks 1 and 2, p < .05, which did not differ (See Table 2). Additionally, there was a significant effect of noise on mood, F(1, 32) = 7.53, p < .05, ηp2 = .19 (see Figure 5.a.), with the noise group endorsing lower mood across all three blocks (See Table 2). There were no significant effects related to the dominance manikin (p > .05; See Figure 4.C), nor were there any significant interactions between block and noise condition (p > .05).
Figure 4.
Change in mood (A), arousal (B), and sense of dominance (C) across blocks in the noise and no-noise conditions. Standard error bars are represented.
Table 2.
Self-reported change in affect by condition
| No Noise |
Noise |
|||
|---|---|---|---|---|
| Affective State | M | SD | M | SD |
| Mood | ||||
| Block 1 | 142.06 | 79.55 | 33.06 | 120.58 |
| Block 2 | 137.94 | 84.64 | 46.76 | 110.64 |
| Block 3 | 60.76 | 127.56 | −10.41 | 131.94 |
| Arousal | ||||
| Block 1 | −177.71 | 140.15 | 107.29 | 131.20 |
| Block 2 | 168.71 | 137.65 | −95.06 | 124.40 |
| Block 3 | −62.18 | 148.53 | −6.71 | 149.27 |
| Dominance | ||||
| Block 1 | 64.06 | 105.32 | 2.18 | 111.70 |
| Block 2 | 27.41 | 88.71 | −1.35 | 81.27 |
| Block 3 | 16.18 | 108.09 | 11.00 | 96.90 |
Discussion
The effects of the affective Posner task were replicated (Rich et al., 2005) in that healthy participants reported reduced mood and increased arousal. In addition, participants demonstrated increased reaction time at the expense of accuracy on block three, which is pulled for by the introduction of the rigged feedback. Although the mean error rate was quite small across blocks, the number of participants committing errors was almost three-fold on Block 3 when compared to Blocks 1 and 2. The presence of scanner noise did not impact task performance, as there were no differences among conditions in accuracy or reaction time. This finding is consistent with previous reports suggesting that the increased attentional load caused by the presence of scanner noise leads to more activation in attentional areas, in turn, preserving performance on relevant tasks (Tomasi et al., 2005; Hommel et al., 2012). Individuals in the noise group did, however, report significantly poorer mood throughout the task. Therefore, although noise led to reduced mood in these participants, it did not affect task performance systematically.
Several limitations are present in the current study. Given the fixed order of the blocks, it is possible that participants habituated to the noise, reducing the impact of the noise across the duration of the task; however, there was still an effect of noise on Block 3 (when any practice effects would likely be maximal). Furthermore, there was no interaction between block and noise condition, suggesting that any practice effects were no different between the two conditions. If a different paradigm were used that had the potential to induce frustration earlier in the testing session, it is possible that the effect of noise would be even greater. A second limitation of the study is that change in emotional experience across blocks was only evaluated with a subjective self-report measure. Physiological data to accompany the self-assessment manikin would provide a more objective assessment of change across the tasks, and while we attempted to include such data acquisition in this study, we were beset by technical problems that invalidated these data. Additionally, a self-assessment baseline was not obtained in the current study, and therefore we cannot be sure that there was not a difference between groups before the noise was introduced.
We also acknowledge that this study was conducted with a healthy college population, limiting the generalizability of the results to other populations. In addition, the impact of scanner noise on other areas of cognition also should be investigated. While the scanner noise did not impact performance or reaction time on an attentional task in healthy college students, the findings may differ for memory or spatial processing and in certain patient populations. Another point to consider is that the task used in this study was relatively easy as evidenced by the low error rate. The finding that task performance was preserved in the presence of scanner noise may not generalize to performance on more difficult tasks. Finally, it cannot be inferred that the findings would generalize to research taking place in the scanner. The volume of the noise was not matched to that of the scanner environment, and therefore, we do not know how these findings would translate to the environment and sounds that subjects experience while performing an MRI study. In addition, the study took place outside of the scanner and did not investigate the effects of other aspects of the scanner environment (e.g., supine position and the enclosed environment) on affect and task performance.
To our knowledge, this is the first study to demonstrate that the affective Posner task successfully induces poorer mood and increased arousal (consistent with increased frustration) and leads to changes in task performance in healthy adults. This study is also novel in that it demonstrated the impact of noise on participants using a well-validated measure on which participants rated their own affective change throughout the task. Given the potential impact of scanner noise on affective state, the findings have important implications for research carried out in the scanner and warrant further investigation. A potential next step is the investigation of how noise might impact performance and affective responses differently in individuals who are more susceptible to potentially stressful stimuli using a within subjects design. In addition, the observed pattern of effects might be different in patients with psychiatric conditions that impact affective functioning or in individuals with certain cognitive disorders. It will also be important to expand this line of research to include the scanner environment, perhaps investigating how the results of the present study compare to those obtained in both a mock and a real scanner environment.
Acknowledgements
Research reported in this publication was supported by the Mental Health Institute of the National Institutes of Health under grant number P50MH077138 (PI: Strakowski) and the Drug Abuse Institute of the National Institutes of Health under grant number K01DA020485 (PI: Eliassen). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
References
- Bradley MM, Lang PJ. Measuring emotion: The self assessment manikin and the semantic differential. Journal of Behavior Therapy and Experimental Psychiatry. 1994;25:49–59.9. doi: 10.1016/0005-7916(94)90063-9. doi:10.1016/0005-7916(94)90063-9. [DOI] [PubMed] [Google Scholar]
- Breier A, Albus M, Pickar D, Zahn TP, Wolkowitz OM, Paul SM. Controllable and uncontrollable stress in humans: Alterations in mood and neuroendocrine and psychophysiological function. American Journal of Psychiatry. 1987;144:1419–1425. doi: 10.1176/ajp.144.11.1419. [DOI] [PubMed] [Google Scholar]
- Harmon-Jones E, Peterson CK. Supine Body Position Reduces Neural Response to Anger Evocation. Psychological Science. 2009;20:1209–1210. doi: 10.1111/j.1467-9280.2009.02416.x. doi:10.1111/j.1467-9280.2009.02416.x. [DOI] [PubMed] [Google Scholar]
- Hommel B, Fischer R, Colzato LS, van den Wildenberg WPM, Cellini C. The effect of fMRI (noise) on cognitive control. Journal of Experimental Psychology. 2012;38:290–301. doi: 10.1037/a0026353. doi: 10.1037/a0026353. [DOI] [PubMed] [Google Scholar]
- Mazzard A, Mazoyer B, Etard O, Tzourio-Mazoyer N, Kosslyn SM, Mellet E. Impact of fMRI acoustic noise on the functional anatomy of visual mental imagery. Journal of Cognitive Neuroscience. 2002;14:172–186. doi: 10.1162/089892902317236821. Retrieved from http://search.proquest.com/docview/71619664?accountid=2909. [DOI] [PubMed] [Google Scholar]
- Melendez J, McCrank E. Anxiety-related reactions associated with magnetic resonance imaging examinations. Journal of the American Medical Association. 1993;270:745–747. doi: 10.1001/jama.1993.03510060091039. doi:10.1001/jama.1993.03510060091039. [DOI] [PubMed] [Google Scholar]
- Moelker A, Pattynama PMT. Acoustic noise concerns in functional magnetic resonance imaging. Human Brain Mapping. 2003;20:121–141. doi: 10.1002/hbm.10134. doi:10.1002/hbm.10134. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Posner MI. Orienting of Attention. Quarterly Journal of Experimental Psychology. 1980;32:3–25. doi: 10.1080/00335558008248231. doi:10.1080/00335558008248231. [DOI] [PubMed] [Google Scholar]
- Rich BA, Schmajuk M, Perez-Edgar KE, Pine DS, Fox NA, Leibenluft E. The Impact of reward, punishment, and frustration on attention in pediatric bipolar disorder. Biological Psychiatry. 2005;58:532–539. doi: 10.1016/j.biopsych.2005.01.006. doi:10.1016/j.biopsych.2005.01.006. [DOI] [PubMed] [Google Scholar]
- Szalma JL, Hancock PA. A meta-analytic review of the effects of noise on performance: Moderating effects of tasks and noise characteristics.. Paper presented at the meeting of the Human Factors and Ergonomics Society; San Francisco, CA.. 2010. [Google Scholar]
- Tessner KD, Walker EF, Hochman K, Hamann S. Cortisol responses of healthy volunteers undergoing magnetic resonance imaging. Human Brain Mapping. 2006;27:889–895. doi: 10.1002/hbm.20229. doi:10.1002/hbn.20229. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Tomasi D, Caparelli ED, Chang L, Ernst T. fMRI-acoustic noise alters brain activation during working memory tasks. NeuroImage. 2005;27:377–386. doi: 10.1016/j.neuroimage.2005.04.010. doi:10.1016/j.neuroimage.2005.04.010. [DOI] [PMC free article] [PubMed] [Google Scholar]