A more sophisticated, neural network analysis of fMRI data has shown that representational content—which can appear, under standard methods of data analysis, to be strictly segregated (e.g., face-vs.-object perception in the ventral temporal lobe)—is actually intermingled and dispersed across a wider region of the cortex. Information encoding appears to depend not on strict localization, but on a combinatorial pattern of variations in the intensity of the neural response across regions once thought to be functionally distinct (Hanson, Matsuka, & Haxby, 2004).
There are also epistemological questions about what it means to correlate any mental state with physiological changes in the brain. And yet, while I consider the so-called "hard problem" of consciousness (Chalmers, 1996) a real barrier to scientific explanation, I do not think it will hinder the progress of cognitive neuroscience generally. The distinction between consciousness and its contents seems paramount. It is true that we do not understand how consciousness emerges from the unconscious activity of neural networks—or even how it could emerge. But we do not need such knowledge to compare states of mind through neuroimaging. To consider one among countless examples from the current literature: neu-roscientists have begun to investigate how envy and schadenfreude are related in neuroanatomical terms. One group found activity in the ACC (anterior cingulate cortex) to be correlated with envy, and the magnitude of signal change was predictive of activity in the striatum (a region often associated with reward) when subjects witnessed those they envied experiencing misfortune (signifying the pleasure of schadenfreude) (Takahashi et al., 2009). This reveals something about the relationship between these mental states that may not be obvious by introspection. The finding that right-sided lesions in the MPFC impair the perception of envy (a negative emotion), while analogous left-sided lesions impair the perception of schadenfreude (a positive emotion) fills in a few more details (Shamay-Tsoory, Tibi-Elhanany, & Aharon-Peretz, 2007)—as there is a wider literature on the lateralization of positive and negative mental states. Granted, the relationship between envy and schadenfreude was somewhat obvious without our learning their neural correlates. But improvements in neuroimaging may one day allow us to understand the relationship between such mental states with great precision. This may deliver conceptual surprises and even personal epiphanies. And if the mental states and capacities most conducive to human well-being are ever understood in terms of their underlying neurophysiology, neuroimaging may become an integral part of any enlightened approach to ethics.
It seems to me that progress on this front does not require that we solve the "hard problem" of consciousness (or that it even admit of a solution). When comparing mental states, the reality of human consciousness is a given. We need not understand how consciousness relates to the behavior of atoms to investigate how emotions like love, compassion, trust, greed, fear, and anger differ (and interact) in neurophysiological terms.
19. Most inputs to cortical dendrites come from neurons in the same region of cortex: very few arrive from other cortical regions or from ascending pathways. For instance, only 5 percent to 10 percent of inputs to layer 4 of visual cortex arrive from the thalamus (R. J. Douglas & Martin, 2007).
20. The apparent (qualified) existence of "grandmother cells" notwithstanding (Quiroga, Reddy, Kreiman, Koch, & Fried, 2005). For a discussion of the limits of traditional "connectionist" accounts of mental representation, see Doumas & Hummel, 2005.
21. These data were subsequently published as Harris, S., Sheth, & Cohen 2008.
22. The post-hoc analysis of neuroimaging data is a limitation of many studies, and in our original paper we acknowledged the importance of distinguishing between results predicted by a specific model of brain function and those that arise in the absence of a prior hypothesis. This caveat notwithstanding, I believe that too much has been made of the distinction between descriptive and hypothesis-driven research in science generally and in neuroscience in particular. There must always be a first experimental observation, and one gets no closer to physical reality by running a follow-up study. To have been the first person to observe blood-flow changes in the right fusiform gyrus in response to visual stimuli depicting faces (Sergent,
Ohta, & MacDonald, 1992)—and to have concluded, on the basis of these data, that this region of cortex plays a role in facial recognition—was a perfectly legitimate instance of scientific induction. Subsequent corroboration of these results increased our collective confidence in this first set of data (Kanwisher, McDermott, & Chun, 1997) but did not constitute an epistemological advance over the first study. All subsequent hypothesis-driven research that has taken the fusiform gyrus as a region of interest derives its increased legitimacy from the descriptive study upon which it is based (or, as has often been the case in neuroscience, from the purely descriptive, clinical literature). If the initial descriptive study was in error, then any hypothesis based on it would be empty (or only accidentally correct); if the initial work was valid, then follow-up work would merely corroborate it and, perhaps, build upon it. The injuries suffered by Phineas Gage and H.M. were inadvertent, descriptive experiments, and the wealth of information learned from these cases—arguably more than was learned from any two experiments in the history of neuroscience—did not suffer for lack of prior hypothesis. Indeed, these clinical observations became the basis of all subsequent hypotheses about the function of the frontal and medial temporal lobes.
23. E. K. Miller & Cohen, 2001; Desimone & Duncan, 1995. While damage to the PFC can result in a range of deficits, the most common is haphazard, inappropriate, and impulsive behavior, along with the inability to acquire new behavioral rules (Bechara, Damasio, & Damasio, 2000). As many parents can attest, the human capacity for self-regulation does not fully develop until after adolescence; this is when the white-matter connections in the PFC finally mature (Sowell, Thompson, Holmes, Jernigan, &Toga, 1999).
24. Spinoza, [1677] 1982.
25. D. T. K. Gilbert, 1991; D. T. K. Gilbert, Douglas, & Malone, 1990; J. E Mitchell, Dodson, & Schacter, 2005.
26. This truth bias may interact with (or underlie) what has come to be known as the "confirmation bias" or "positive test strategy" heuristic in reasoning (Klay-man & Ha, 1987): people tend to seek evidence that confirms an hypothesis rather than evidence that refutes it. This strategy is known to produce frequent reasoning errors. Our bias toward belief may also explain the "illusory-truth effect," where mere exposure to a proposition, even when it was revealed to be false or attributed to an unreliable source, increases the likelihood that it will later be remembered as being true (Begg, Robertson, Gruppuso, Anas, & Needham, 1996; J. P. Mitchell etal., 2005).
27. This was due to a greater decrease in signal during disbelief trials than during belief trials. This region of the brain is known to have a high level of resting-state activity and to show reduced activity compared to baseline for a wide variety of cognitive tasks (Raichle et al., 2001).
28. Bechara et al., 2000. The MPFC is also activated by reasoning tasks that incorporate high emotional salience (Goel & Dolan, 2003b; Northoff et al., 2004). Individuals with MPFC lesions test normally on a variety of executive function tasks but often fail to integrate appropriate emotional responses into their reasoning about the world. They also fail to habituate normally to unpleasant somatosensory stimuli (Rule, Shimamura, & Knight, 2002). The circuitry in this region that links decision making to emotions seems rather specific, as MPFC lesions do not disrupt fear conditioning or the normal modulation of memory by emotionally charged stimuli (Bechara et al., 2000). While reasoning appropriately about the likely consequences of their actions, these persons seem unable to feel the difference between good and bad choices.
29. Hornak et al., 2004; O'Doherty, Kringelbach, Rolls, Hornak, & Andrews, 2001.
30. Matsumoto&Tanaka, 2004.
31. Schnider,2001.
32. Northoffetal.,2006.
33. Kelleyetal.,2002.
34. When compared with both belief and uncertainty, disbelief was a
ssociated in our study with bilateral activation of the anterior insula, a primary region for the sensation of taste (Faurion, Cerf, Le Bihan, & Pillias, 1998; O'Doherty, Rolls, Francis, Bowtell, & McGlone, 2001). This area is widely rhought to be involved with negatively valenced feelings like disgust (Royet, Plailly, Delon-Martin, Kareken, & Segebarth, 2003; Wicker et al, 2003), harm avoidance (Paulus, Rogal-sky, Simmons, Feinstein, & Stein, 2003), and the expectation of loss in decision tasks (Kuhnen & Knutson, 2005). The anterior insula has also been linked to pain perception (Wager et al., 2004) and even to the perception of pain in others (T. Singer et al., 2004). The frequent association between activity in the anterior insula and negative affect appears to make at least provisional sense of the emotional tone of disbelief.
While disgust is tegularly classed as a primary human emotion, infants and toddlers do not appear to feel it (Bloom, 2004, p. 155). This would account for some of their more arresting displays of incivility. Interestingly, people suffering from Huntington's disease, as well as presymptomatic carriers of the HD allele, exhibit reduced feelings of disgust and are generally unable to recognize the emotion in others (Calder, Keane, Manes, Antoun, & Young, 2000; Gray, Young, Barker, Curtis, & Gibson, 1997; Halligan, 1998; Hayes, Stevenson, & Coltheart, 2007; I.J. Mitchell, Heims, Neville, & Rickatds, 2005; Sprengelmeyer, Schroeder, Young, & Epplen, 2006). The recognition deficit has been correlated with reduced activity in the anterior insula (Hennenlotter et al., 2004; Kipps, Duggins, McCusker, & Calder, 2007)—though other work has found that HD patients and carriers are impaired in processing a range of (predominantly negative) emotions: including disgust, anger, fear, sadness, and surprise (Henley et al., 2008; Johnson et al., 2007; Snowden et al., 2008).
We must be careful not to draw too strong a connection between disbelief and disgust (or any other mental state) on the basis of these data. While a connection between these states of mind seems intuitively plausible, equating disbelief with disgust represents a "reverse inference" of a sort known to be problematic in the field of neuroimaging (Poldrack, 2006). One cannot reliably infer the presence of a mental state on the basis of brain data alone, unless the brain regions in question are known to be truly selective for a single mental state. If it were known, for instance, that the anterior insulae were active if and only if subjects experienced disgust, then we could draw quite a strong inference about the role of disgust in disbelief. But there are very few regions of the brain whose function is so selective as to justify inferences of this kind. The anterior insula, for instance, appears to be involved in a wide range of neutral/positive states—including time perception, music appreciation, self-recognition, and smiling (A. D. Craig, 2009).
And there may also be many forms of disgust: While subjects tend to rate a wide range of stimuli as equivalently "disgusting," one group found that disgust associated with pathogen-related acts, social-sexual acts (e.g., incest), and nonsexual moral violations activated different (but overlapping) brain networks (J. S. Borg, Lieberman, & Kiehl, 2008). To further complicate matters, they did not find the insula implicated in any of this disgust processing, with the exception of the subjects' response to incest. This group is not alone in suggesting that the insula may not be selective for disgust and may be more generally sensitive to other factors, including self-monitoring and emotional salience. As the authors note, the difficulty in interpreting these results is compounded by the fact that their subjects were engaged in a memory task and not required to explicitly evaluate how disgusting a stimulus was until after the scanning session. This may have selected against insular activity; at least one other study suggests that the insula may only be preferentially active in response to attended stimuli (Anderson, Christoff, Panitz, De Rosa, & Gabrieli, 2003).
35. These results seem to pull the rug out from under one widely subscribed view in moral philosophy, generally described as "non-cognitivism." Non-cognitivists hold that moral claims lack propositional content and, therefore, do not express genuine beliefs about the world. Unfortunately for this view, our brains appear to be unaware of this breakthrough in metaethics: we seem to accept the truth of moral assertions in the same way as we accept any other statements of fact.
In this first experiment on belief, we also analyzed the brain's response to uncertainty: the mental state in which the truth value of a proposition cannot be judged. Not knowing what one believes to be true— Is the hotel north of Main Street, or south of Main Street? Was he talking to me, or to the man behind me? —has obvious behavioral/emotional consequences. Uncertainty prevents the link between thought and subsequent behavior/emotion from forming. It can be distinguished readily from belief and disbelief in this regard, because in the latter states, the mind has settled upon a specific, actionable representation of the world. The results of our study suggest two mechanisms that might account for this difference.
The contrasts— uncertainty minus belief and uncertainty minus disbelief- —yielded signal in the anterior cingulate cortex (ACC). This region of the brain has been widely implicated in error detection (Schall, Stuphorn, & Brown, 2002) and response conflict (Gehring & Fencsik, 2001), and it regularly responds to increases in cognitive load and interference (Bunge, Ochsner, Desmond, Glover, & Gabrieli, 2001). It has also been shown to play a role in the perception of pain (Coghill, McHaffie, & Yen, 2003).
The opposite contrasts, belief minus uncertainty and disbelief minus uncertainty, showed increased signal in the caudate nucleus, which is part of the basal ganglia. One of the primary functions of the basal ganglia is to provide a route by which cortical association areas can influence motor action. The caudate has displayed context-specific, anticipatory, and reward-related activity in a variety of animal studies (Mink, 1996) and has been associated with cognitive planning in humans (Monchi, Petrides, Strafella, Worsley, & Doyon, 2006). It has also been shown to respond to feedback in both reasoning and guessing tasks when compared to the same tasks without feedback (Elliott, Frith, & Dolan, 1997).
In cognitive terms, one of the principal features of feedback is that it systematically removes uncertainty. The fact that both belief and disbelief showed highly localized signal changes in the caudate, when compared to uncertainty, appears to implicate basal ganglia circuits in the acceptance or rejection of linguistic representations of the world. Delgado et al. showed that the caudate response to feedback can be modulated by prior expectations (Delgado, Frank, & Phelps, 2005). In a trust game played with three hypothetical partners (neutral, bad, and good), they found that the caudate responded strongly to violations of trust by a neutral partner, to a lesser degree with a bad partner, but not at all when the partner was assumed to be morally good. On their account, it seems that the assumption of moral goodness in a partner led subjects to ignore or discount feedback. This result seems convergent with our own: one might say that subjects in their study were uncertain of what to conclude when a trusted collaborator failed to cooperate.
The ACC and the caudate display an unusual degree of connectivity, as the surgical lesioning of the ACC (a procedure known as a cingulotomy) causes atrophy of the caudate, and the disruption of this pathway is thought to be the basis of the procedure's effect in treating conditions like obsessive-compulsive disotder (Rauch er al., 2000; Rauch et al., 2001).
There are, however, different types of uncertainty. For instance, there is a difference between expected uncertainty—where one knows that one's observations are unreliable—and unexpected uncertainty, where something in the environment indicates that things are not as they seem. The difference between these two modes of cognition has been analyzed within a Bayesian statistical framework in terms of their underlying neurophysiology. It appears that expected uncertainty is largely mediated by acetylcholine and unexpected uncertainty by norepinephrine (Yu & Dayan, 2005). Behavioral economists sometimes distinguish between "risk" and "ambiguity": the former being a condition where probability can be assessed, as in a game of roulette, the latter being rhe uncertainty borne of missing information.
People are generally more willing to take even very low-probability bets in a condition of risk than they are to act in a condition of missing information. One group found that ambiguity was negatively correlated with activity in the dorsal striatum (caudate/putamen) (Hsu, Bhatt, Adolphs, Tranel, & Camerer, 2005). This result fits very well with our own, as the uncertainty provoked by our stimuli would have taken the form of "ambiguity" rather than "risk."
36. There are many factors that bias our judgment, including: arbitrary anchors on estimates of quantity, availability biases on estimates of frequency, insensitivity to the prior probability of outcomes, misconceptions of randomness, nonregressive predictions, insensitivity to sample size, illusory correlations, overconfidence, valuing of worthless evidence, hindsight bias, confirmation bias, biases based on ease of imaginability, as well as other nonnormative modes of thinking. See Baron, 2008; J. S. B.T. Evans, 2005; Kahneman, 2003; Kahneman, Krueger, Schkade, Schwarz, & Stone, 2006; Kahneman, Slovic, &Tversky, 1982; Kahneman &Tversky, 1996; Stanovich & West, 2000; Tversky & Kahneman, 1974.