The Emotional Foundations of Personality

by Kenneth L Davis

  To summarize, when the threat is more distant, cortical activity is more prominent, and when threat is near with increased dread and decreased confidence of escape, PAG activity is more prominent. “From an evolutionary viewpoint, higher cortical systems control behavior when the degree of threat is appraised as not-life-endangering. . . . At extreme levels of threat, the PAG may in turn inhibit more complex control processes when a fast and indeed obligatory response is required” (Mobbs et al., 2007, p. 1082). So, consistent with cross-species affective neuroscience research, and continuing the theme of this chapter, low levels of fear are accompanied by cortical arousal and corresponding cognitive threat assessment, but as the level of threat mounts, the PAG in combination with other subcortical regions inhibit cortical activity, exert increasing influence over the brain, and provides rapid reactions based on evolutionary affective memories, along with one’s personal higher autobiographical history, to prepare and guide the body for survival.

  Mobbs et al. (2009) repeated their simulated predator and human prey experiment and largely replicated their results. In this study, they also measured subjects’ skin conductance levels (a physiological measure of anxiety) during the experiment. After subjects emerged from the scanner, researchers also asked them to rate how much anxiety they felt during the various phases of the simulation, as well as how much panic they felt during the predator encounters. These measures allowed researchers to verify that anxiety onset occurred when an encounter with the predator was signaled and that skin conductance levels, anxiety, and panic were all at their highest levels when actually being chased by the predator in the high-danger condition, when midbrain PAG activity also peaked, rather than in the low danger condition. It is relevant to note that we would prefer to reserve the term panic for use with the PANIC/Sadness brain system rather than an extreme expression of the FEAR system. This clearly highlights the need for a different lexicon for primary-process emotions, which we try to achieve with the convention of full-capitalization of primal emotional terms). In any case, we will continue to use Mobb’s terminology.

  In this replication, researchers also measured the number of button-press errors (e.g., accidentally guiding their computer icon into the wall of the maze), which likewise peaked during the high-danger chase and correlated with self-rated panic levels. Indeed, they “found that midbrain [PAG] activity increased with the amount of panic-related locomotor errors,” which was consistent with “chemical stimulation of the rodent dorsolateral PAG eliciting uncoordinated panic-like behaviors” (Mobbs et al., 2009, p. 12,241). (Beside FEAR, the separation-distress PANIC response is also well represented in the PAG.) Further, the increase of “all thumbs” uncoordinated fine motor responding that Mobbs and colleagues observed is consistent with the idea that the frontal cortex motor planning functions are inhibited (or disrupted) during high levels of threat that require faster, more ancient, evolutionarily conserved escape responses, while on the other end of the affective seesaw, the prefrontal cortex can exert regulatory inhibition on subcortical regions when lower threat levels allow for more carefully planned and cognitively coordinated escape strategies.

  BRIDGING THE HUMAN AND ANIMAL PAG RESEARCH

  Studies such as those by Mobbs and colleagues have encouraged other researchers to use fMRI procedures to focus more on the role of the subcortical PAG in the experience of emotions. Indeed, Buhle et al. (2013) took on the challenge of investigating whether they could replicate animal research and elicit comparable PAG activation in human subjects using negative emotional responses to pain or negative emotional responses while viewing aversive photographs (taken from the International Affective Picture System of Lang, Bradley, & Cuthbert, 2008).

  All participants were subjected to high and low thermal pain as well as aversive or neutral cognitive images, and then were asked to rate how negatively they felt about the stimulus. The participants consistently reported that they experienced greater negative affect from high heat than from low heat, and from the aversive images than from the neutral images. Furthermore, the ratings for the painful heat and the aversive images did not significantly differ. Importantly, Buhle et al. noted that “Whole brain contrasts of both high vs. low pain and negative vs. neutral image viewing revealed activity in the PAG” and that “the activity did not reliably differ between the conditions [pain or aversive image]” (2013, p. 611).

  As an additional check on their results, Buhle et al. (2013) identified eight independent human studies, four examining responses to high or low pain and four studying responses to negative versus neutral images. In each of the eight independent data sets, whole brain analyses identified activity in the PAG. They concluded that, combined with their own results, “these results support the hypothesis that PAG plays an important role in human negative affect, in line with previous evidence from research in animals” (p. 612). In sum, beneath our “crowning glory” of neocortex, we are like other mammals below our “thinking caps.” In this context, it is important to recognize that practically all mammalian cortical functions (e.g., including vision) are probably learned (Sur & Rubinstein, 2005) rather than tightly programmed by brain evolution.

  SHARPER FOCUS ON THE HUMAN PAG

  Apart from examining the PAG as a whole, animal research has shown that there are distinct subregions of the PAG involving specific affective processes (Bandler & Shipley, 1994). Indeed, Sapute et al. (2013) set out to determine whether PAG subregions could be identified in humans, thus further supporting the homologies between human and animal emotions. Yet, the PAG is difficult to accurately isolate, let alone subdivide, using typical brain scanning procedures. This is in part due to the small size of the PAG—about 10 mm long (three-eighths of an inch), with a diameter of about 6 mm (less than a quarter inch) in humans. Further adding to difficulties imaging the PAG, its structure is shaped like a sleeve or hollow cylinder surrounding the cerebral aqueduct, such that the inner half of the PAG’s diameter is the cerebral aqueduct, a part of the brain that, as its name implies, ensures the flow of cerebrospinal fluid, mainly downward, from the rest of the brain. During fMRI scanning, strong signals from the fluid in the cerebral aqueduct can interfere with and mask signals from the PAG. Sapute et al. (2013) addressed these problems by using an fMRI procedure incorporating an exceptionally strong 7-tesla magnet, which could provide higher scanning resolution—down to 0.75 mm—than typical fMRI equipment.

  Sapute’s group elicited emotions in their eleven scanning participants by showing them either neutral or highly aversive images that were related to threat, harm, and loss (again, taken from the International Affective Picture System). After having their brains scanned while viewing a set of images, participants were asked to report their emotional response to the images with five separate emotional labels: “Activated” (for arousal), “Angry,” “Disgusted,” “Sad,” and “Scared” (for fear). The emotional labels were always presented in random order and were rated on a five-point, low-to-high scale.

  An initial analysis showed that overall activity in the PAG was greater when subjects were viewing highly aversive compared to neutral images. However, an exploratory factor analysis of the high-resolution scanning results from their 7-tesla fMRI, along with subjects’ self-rated emotional experiences when viewing the aversive versus neutral images, yielded three factors representing three different PAG subregions, with each of the subregions corresponding with a different emotional experience: They reported not only having “observed definitive activation in the human PAG” but also that “segmenting the PAG into both radial and longitudinal subregions illustrated that activity during negative affect was not diffuse but was concentrated along a spiral pattern from ventrolateral caudal PAG to lateral and dorsomedial rostral PAG. This [spiral-like] pattern mirrors functional and structural observations in nonhuman animals” (Sapute et al., 2013, p. 17,104). Further, spiraling around the central aqueduct from caudal to rostral (tail to head), the three PAG subregions generally corresponded to (1) disgust, arousa
l, and fear; (2) anger; and (3) sadness.

  In short, Sapute and colleagues provided robust evidence in support of evolutionarily conserved mammalian brain homologies—from mice to men, so to speak; they had demonstrated that ultra-high-resolution fMRI procedures could be used to explore the functional architecture of the PAG, an approach that could perhaps be extended to other midbrain regions, which have so far been largely ignored in brain imaging studies of human emotion. Still, it is gratifying to see such clear functional continuities across all mammalian species that have been studied. It reinforces the conclusion that we all share a variety of evolutionarily conserved basic emotions.

  REVIEWS OF NEUROTICISM AND HUMAN BRAIN IMAGING: META-ANALYSIS

  Meta-analyses allow researchers to statistically combine the data from many published scientific studies and obtain a collective result that may be more valid than any of the individual studies alone. The underlying idea is that, while a single study might not report valid results because of various procedural problems or sampling errors, when the data of multiple studies are pooled something closer to a true picture is likely to emerge.

  One such meta-analysis reported by Servaas et al. (2013) identified eighteen studies published from 2001 to 2011 that provided fMRI brain imaging data and self-report measures of neuroticism (also sometimes called negative emotionality or low emotional stability) using psychologically healthy subjects as participants. Their concern was that individual studies attempting to use neuroimaging to identify neurobiological correlates of neuroticism had yielded inconsistent findings. They hoped that merging these generally similar studies into a single meta-analysis would reveal more consistent data patterns and thus better validate the different roles of various brain regions than the individual studies.

  Remarkably, none of the brain regions the Servaas group identified as being positively correlated with neuroticism were subcortical. Indeed, there was no mention of periaqueductal gray (PAG) activity. In search for the clearest associations with neuroticism, they identified three general brain regions that positively correlated with self-report measures of neuroticism. The first was the left parahippocampal gyrus, along with closely adjacent areas, which were primarily associated with fear-conditioning studies. The second and third areas were the left superior frontal gyrus and the dorsal and ventral regions of the right middle cingulate gyrus. These latter two areas were associated with general emotional processing, such as viewing negative emotional facial expressions, categorizing emotional words, and choice tasks resulting in the relative loss or gain of small amounts of money ($4 or less). Why the disjunction with studies monitoring immediate emotional feelings?

  We would suggest that personality is ultimately an acquired result of how past primary emotional arousals have helped construct diverse brain action systems that mediate emotional activity. In the Servaas meta-analysis, each of the three general personality-relevant areas had previously been associated with cognitive emotional processing. The parahippocampal gyrus had been shown to interact with the amygdala during the encoding of negative film clips (Kilpatrick & Cahill, 2004) and in the recall of negative words (Thomaes et al., 2009). The superior frontal gyrus had been shown to be involved with maintaining human self-awareness (Goldberg, Harel, & Malach, 2006) and had been associated with general cognitive control and perhaps especially modulating the current emotional state (Frank et al., 2014). The dorsal and ventral regions of the cingulate gyrus seemed to be important in regulating the balance between external and internal attentional factors (Leech & Sharp, 2014).

  Another reason that Servaas et al. (2013) did not identify any subcortical associations with neuroticism is possibly that the NEO PI-R Neuroticism scale was used to measure neuroticism in twelve of the studies they reviewed, a scale that mainly deals with a tertiary cortical cognitive appraisal of negative emotion. Items like “I have fewer fears than most people” and “Frightening thoughts sometimes come into my head” may require more cognitive reflection than a direct assessment of how the self-rater feels at the moment. Also, items like “I feel I am capable of coping with most of my problems” and “I can handle myself pretty well in a crisis” not only entail cognitive reflection but also are very general and may not tap into the patterns of specific primary emotional tendencies that have guided personality development.

  A review by Montag et al. (2013) has suggested that a brain-based personality assessment such as the Affective Neuroscience Personality Scales (ANPS) might be better suited for parsing emotion-related brain regions. The ANPS was designed to address the primary-process negative emotions, namely, RAGE/Anger, FEAR, and PANIC/Sadness, and uses items that more directly tap into a self-rater’s affective experience rather than relying on more general affective judgments. It is no coincidence that the ANPS negative emotionality scales target the three primary emotions Sapute et al. (2013) linked to specific regions of the periaqueductal gray using fMRI imaging: fear, anger, and sadness.

  Yet another possibility is that the tasks used in the eighteen studies reviewed by Servaas et al. (2013), such as viewing negative emotional facial expressions and categorizing emotional words, are more cognitively oriented and may not strongly engage strong emotional feelings. Those tasks may be more similar to the early stage of the predator task used in the fMRI study by Mobbs et al. (2009), when there was no imminent danger and the participant had not yet encountered the predator.

  In any event, the overall meta-analytic results by Servaas et al. (2013) seem more reflective of a cognitive neuroscience approach to emotions, which is inclined to look for sources of emotional brain activity in the human cortex than in subcortical regions such as the PAG originally identified in animal research, which is only recently becoming accepted as a key region for the experience of human emotion. Thus, we need to see human personality in part as an “emergent process” of how basic emotional arousals have guided the life trajectories of individual human beings. Personality may reflect how the ancient primal affective tools for living guide how one has learned to be a specific type of person in a specific environment.

  EXTENDING FINDINGS WITH MORE META-ANALYSIS

  Another meta-analysis by Adina Mincic (2015) took advantage of a very active new research field, examining fifty-seven studies relating negative emotionality to brain activity and brain structure sizes, with most studies reporting cortical gray matter differences. A predominant finding was that higher negative emotionality correlated with reduced gray matter volume in the left orbitofrontal cortex (OFC)—sometimes included in the ventromedial prefrontal cortex—a region positioned just above and behind the eye sockets in humans.

  Another prominent finding was greater gray matter volume in the left amygdala for participants with higher negative emotionality. Mincic (2015) also found evidence for increased volume in the hippocampus and the parahippocampal gyrus associated with higher negative emotionality scores. She only found a few studies focusing on the cingulate cortex—an evolutionarily older cortical region lying immediately above the corpus callosum—but focused on the anterior cingulate cortex ACC rather than the posterior cingulate cortex, with the reduced OFC volume extending into the rostral (very front portion) of ACC, which lies adjacent to and immediately behind the OFC. In addition, Mincic (2015) found neuroanatomical evidence for decreased volume in the left uncinate fasciculus, a Latin name for a nerve tract connecting frontal areas such as the OFC to the amygdala/hippocampal regions. All of this “supports the idea that the diminished grey matter in OFC/ACC and white matter integrity of the uncinate fasciculus may represent a structural phenotype of the NE [Negative Emotionality]-related personality traits” (p. 110). Such a conclusion highlights the top-down regulation of emotional arousal in maintaining emotional balance and well-being. However, these studies are still missing the critical midbrain emotional foundational areas, which are so apparent in the work of Damasio, Mobbs, and Sapute, as already highlighted.

  Another relevant concern is the heterogeneity of results in the literatu
re covered by these two meta-analyses. For starters, the only result the two had in common was a positive correlation in the parahippocampal region with negative emotionality. Mincic (2015) pointed out that Servaas et al. (2013) had found increased activity in the parahippocampal gyrus during negative emotional processing but, surprisingly, not in the amygdala. However, regarding Mincic’s most consistent finding of reduced OFC volume associated with higher negative emotionality, she pointed out that of thirty studies investigating this particular relationship, twelve did not find this association, with two actually reporting opposite results. Similar disparities were evident for each of her conclusions.

  Why is there such inconsistency in the brain imaging literature? Why did these two meta-analyses report such inconsistent results, in stark contrast to the animal studies? There are likely many reasons, including variations in the size and anatomy of human brains, the frequent use of low-resolution scanners, difficulty in the use of fMRI to identify activity in midbrain regions such as the PAG where neurons fire quite slowly relative to higher brain regions, and the fact that many studies do not even target midbrain regions. However, we may need to probe deeper into the reason for these disparities: The midbrain regions and structures like the hypothalamus have strongly inherited genetic/anatomical foundations, while the cortical regions are not as tightly genetically programmed but, rather, acquire their functions much more through individual learning and development, providing another source of cortical anatomical heterogeneity.