Jean-Marc Fellous & Michael Arbib (eds.), Who Needs Emotions?, Oxford U. P. 2005. Chapter 4. [Comments in square brackets are my own thoughts.] Ch.4: TOWARD BASIC PRINCIPLES FOR EMOTIONAL PROCESSING (p. 79-202) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Jean-Marc Fellous and Joseph Ledoux This chapter gives a detailed account of neural systems and pathways involved in fearful reactions to a stimulus. (The stimulus might be a tone that presages a foot-shock administered to a rat; or just placement of the rat in a cage that it has learned to associate with the foot-shock experiments; or, for a person, encounter with a dangerous animal; or for PTSD patients, any situation that they associate with past traumatic events.) Their methodological point is that neuroscientists studying emotion should - study processes, rather than trying to understand particular brain regions in isolation; - divide and conquer: don't try to study "emotions" generically, because the various emotions involve partly separate pathways and processes; hence the particular focus on fear here; - formulate conjectures about other emotions by making informed guesses about what "circuits" these emotions might involve, by analogy with emotions already better undestood; they follow their own advice and offer a conjectural analysis of love (pair bonding). One pleasing aspect of the chapter, besides its impressive level of scholarship, is that while they regard a focus on subjective experience as "shackles" in the scientific study of the brain, they nonetheless take the puzzle of subjective experience very seriously, and make some speculative remarks along the way. [In this respect they differ from Baars, for example, and even from Ortony et al. and Rolls, who largely ignore the matter.] Also one of the most striking messages in the chapter [for me] is its apparent implication that only humans have a working memory. Given the central role they assign to working memory in consciousness, it would seem to follow that animals have nothing like our kind of consciousness (they speculate p.101 that they may have a more specific, more passive sort of consciousness). [But since they mostly study rat brains, and since they are first and foremost neuroscientists, they may have the bias I mentioned in connection with Rolls... When they discuss contextual fear conditioning, their account of this seems to involve bringing together long-term memories with more immediate perceptions -- but isn't this supposed to happen in working memory?] History of the study of emotions ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ They mention the early interest in emotion at the end of the 19th century and in the first half of the 20th century, and its decline in the latter half. They attribute this to - the emergence of cognitive science and AI, both of which tended to focus on thinking rather than emotion - the sense (in mid-century) that emotion was rather well- understood in terms of the "limbic system", whereas research on understanding of thinking had lagged - the cloud of subjectivist suspicion that hung for many decades over the study of emotion. But the idea of the limbic system was vague, inconsistent, and without real explanatory power (i.e., emotion was in fact NOT well-understood at all). They way forward, they say, was and remains to set aside the traditional preoccupation with subjective experience -- as cognitive science and neuroscience learned to do, and focus on process, not structure alone. Fear circuits (p.86-) ~~~~~~~~~~~~~ They focus on conditioning based on a tone (the conditioned stimulus, CS) followed by a shock (the unconditioned stimulus, US), and describe two pathways in the processing of a sound: - a path from the auditory nerves to the thalamus directly to the LA (lateral nucleus) and then to the CE (central nucleus) of the amygdala; and - a path from the auditory nerves to the thalamus TO THE AUDITORY CORTEX (which interprets complex sounds) and then to the LA & CE. {From the amygdala, pathways lead to the brainstem which initiates fear reactions.) Apparently conditioning by CS-US pairs can involve either or both pathways, though the latter seems important for more complex sounds. [Didn't Rolls say that the amygdala involves +ve/-ve evaluation of inputs, and that *associations* reside in the orbitofrontal cortex?? Hmm... there *is* mention on p. 99 of the "orbital prefrontal cortex", also called the ventral prefrontal cortex; there they say this also receives signals from the amygdala and hippocampus, as well as being interconnected with the medial prefrontal cortex (anterior cingulate).] Contextual stimuli (the feared cage) involve a third pathway going from the sensory system to the hippocampus (which is implicated strongly in memory formation) and thence to another pair of nuclei in the amygdala (the basal, B and accessory basal, AB nuclei) The amagdala also sends signals to the medial temporal cortex (see below) enabling the storage of "explicit" emotional memories. Anxiety disorders and PTSD in humans seem to have the same sort of conditioning basis. There's also a discussion (p. 94) how the above pathways become feedback loops; e.g., signals arriving at the brainstem cause general arousal, which then tends to reinforce the activity of already active working memory contents and the sensory cortex, which in turn feed into the amygdala, etc. As one site involved in working memory, they mention "medial prefrontal cortex" (p.99), and the amygdala connects to this as well, and can thereby influence what we attend to. Interestingly. "for the amygdala to respond to fear, the prefrontal region has to be shut down". So in principle you can shut out the fear response by thinking hard about the situation?! According to p.96. the amygdala is essential to storing "implicit" emotional memories (i.e., not consciously accessible?) while the medial temporal lobe is essential to storing "explict" emotional memories (consciously accessible?). [Recall the crucial role of the temporal lobe in making sense of what we see, as well.] p.97 they talk about similarities between computational and neuroscientific models of cognition -- the idea of working memory as a set of temporary storage buffers (different ones for hearing, images, etc.); and also the idea of working memory as a "workspace" or "blackboard", in computational terms. [Quite surprisingly they don't mention Baars in this connection!!] They think that consciousness corresponds to particular representations in working memory, though they concede they don't know why these should lead to subjective experience. The emotional component of a conscious experience is presumably also represented in working memory (p.98), but it is unknown in what form. But the emotional aspect of a representation is produced by the amygdala: "Amygdala activation,... turns a plain perceptual experience into a fearful one." See earlier bracketed remarks about the orbital prefrontal cortex. On p.99, the authors say, "Humans with orbital cortex damage become oblivious to social and emotional cues, have poor decision- making abilities, and may exhibit sociopathic behavior." [Not quite the same role Rolls assigned to the orbitofrontal cortex, viz., encoding of associations between situations/events (stimuli) and rewards/punishments. But clearly such associations are crucial for sensible decision-making!] p.100 they also mention feedback effects via hormonal release (stress hormones, in the case of fear.) Love (p.101-4) ~~~~~~~~~~~~~~ Studies of prairie voles (that form long-lasting monogamous pair bonds) vs. montane voles (that don't) can shed light on this subject. [Recall Baars' emphasis on "contrastive phenomenology"!] It seems the hormone oxytocin is crucial for female (but not male) bonding, while vasopressin is crucial in male (but not female) bonding. Humans have these hormones too, and they are released both into the blood and the brain (i.e., they also act as neurotransmitters). [The current Wikipedia entry implicates these hormones in memory, but not love... But a Mar 21/12 PBS program about pleasure and pain indicated that oxytocin levels rise in both the bride and the groom in a wedding ceremony -- as well as (to a smaller degree) in their close relatives); it is also implicated in parent-baby bonding (both parent and baby experience increased oxytocin release when handling the baby).] They suggest a speculative picture of what happens when you see a loved one. They first assume (uncontroversially, I imagine) that the image is processed as usual in the vision cortex, with signals then proceeding both to the prefrontal cortex and to the temporal lobe. Then they posit activation of subcortical "attachment" centers (apparently, in the medial and posterior nuclei of the amygdala) that they suggest play a similar role in love as the the lateral and central nuclei in fear. [This is consistent with Rolls' observation that both rewards and punishers are encoded in the amygdala -- in fact, he mentioned punishers only in passing. Recall the mention of direct stimulation via electrodes.] They suggest that as in fear, there are connections from these centers to the prefrontal cortex, which then affects working memory. Somehow all these effects are integrated into the working memory, and voila! a loving feeling. Conclusion (p.105) ~~~~~~~~~~~~~~~~~ They list features that they think are generally involved in emotion processing - ancient, robust subcortical circuits; - somewhat independent cognitive circuits; - formation of relatively vivid emotional memories; - 2 routes for processing: thalamus --> amygdala (fast), and thalamus --> cortex --> amygdala (slower); - separate input nuclei in any emotional system (in the amygdala), one for simple inputs (in fear, the LA nucleus), and one for complex ones (in fear, the B/AB nuclei); - a third nucleus (in fear, CE in amygdala) triggers response (involving multiple systems) They suggest that this should be a useful guide for designing emotional robots. [Hmm...]