Coan and Allen (2004) estimated that over 70 published EEG studies have examined frontal asymmetries in emotion. As EEG methods have advanced, so has knowledge of the dynamics of these asymmetries. For example, analyses of spectral activity across frontal electrode sites have shown that important differences in left and right frontal activity in depression and anxiety are relative rather than absolute (Bell, Schwartz, Hardin, Baldwin, & Kline, 1998; Bruder et al., 1997; Gotlib, Ranganath, & Rosenfeld, 1998). Although data from EEG studies using few electrodes have generally supported the finding that individuals with depression and anxiety show relatively less left frontal activity, the limited spatial resolution of this methodology constrains the ability to localize this activity within frontal cortex. Recent years have seen dramatic improvements in the spatial resolution of EEG, due primarily to increased electrode densities, more common availability of structural MRIs, and improved source localization techniques. However, many previous studies of brain asymmetry have not capitalized on these advances. Very few of these EEG studies have reliably identified electrical signals from deep frontal regions (e.g., orbital and medial frontal cortex), as the observed scalp distribution of signals from these regions is often difficult to disambiguate from signals closer to the scalp (Davidson, 2004). The incorporation of structural MRI and fMRI information can greatly improve our ability to localize EEG signal in deep structures, but to date few studies of depression and anxiety have capitalized on this combined approach.
PET and fMRI have been used extensively in recent years to localize specific regions related to emotion, depression, and anxiety (for reviews, see Murphy, Nimmo-Smith, & Lawrence, 2003; Wager et al., 2003). Studies using these techniques can provide somewhat better localization information than EEG, particularly for deep structures. It is thus striking that virtually no PET or fMRI studies have robustly replicated the EEG asymmetry findings (Wager et al., 2003). As has been pointed out in recent articles, this represents a significant problem, calling into question either the asymmetry itself or the methods used to measure it (Canli, 1999; Davidson, 1998, 2002; Davidson & Irwin, 1999; Herrington et al., 2005).
In their recent meta-analysis, Wager and colleagues (2003) concluded that there was only "limited support for valence-specific lateralization of emotional activity in frontal cortex" (p. 513) in the hemodynamic literature. When analyzing studies designed to assess brain activity during approach-withdrawal and positive-negative affective states, they found only a trend toward increased activity in left versus right frontal cortex for approach versus withdrawal, and no effect of hemisphere for positive compared to negative stimuli or states. However, an examination of their methods calls this null finding into question. Of critical importance is that most of the studies used in their meta-analysis did not actually directly test laterality effects. In an effort to compensate for this critical shortcoming, Wager and colleagues used a form of conjunction analysis1 in order to infer laterality effects in the studies they examined. As explained below, conjunction analyses are frequently insensitive to laterality effects in PET and fMRI; thus reliance on this approach limits the conclusions that can be drawn. This criticism also applies to a meta-analysis by Murphy and colleagues (2003), who also examined PET and fMRI studies to test the anterior asymmetry hypothesis in nonclinical populations. They concluded that theories of anterior asymmetries "may be too coarse, in terms of both their neural underpinnings and the aspect of emotion under consideration" (p. 227). However, in the absence of hemodynamic imaging studies using robust asymmetry analyses, the conclusions of these meta-analyses cannot be accepted with confidence.
Because of the failure of hemodynamic methods to replicate the EEG results, it remains to be seen which specific areas of prefrontal cortex are driving the EEG laterality effects. Some researchers have argued that dorsolateral prefrontal cortex (DLPFC) is the key region relating frontal EEG lateralization to emotional valence and motivation (Davidson, 2004; Herrington et al., 2005). However, it is quite possible that, due to the relative ease with which EEG can detect signals from regions near the scalp surface, electrical activity from DLPFC may overshadow important lateralized activity in deeper structures. For example, in addition to DLPFC, ventromedial prefrontal cortex (VMPFC) and regions of anterior cingulate cortex play important roles in emotion (Davidson & Irwin, 1999). It is possible that these regions also show lateralized activity patterns that explain important variance in emotion and motivation.
Because of the relative consistency of EEG techniques and findings in this area over the past two decades, it can be argued that limitations in hemodynamic imaging paradigms, techniques, and analyses are central to the failure of hemodynamic imaging to replicate EEG findings regarding frontal lateralization. The following sections examine theoretical and methodological areas where hemodynamic imaging studies may be falling short.
Was this article helpful?