Functional network for learning about bad, wrong or suboptimal choices,
deeds and consequences – directed to inhibitory avoidance and self-control,
via suppression of dopamine and serotonin signaling.
Karin Vadovičová, Roberto Gasparotti
Abstract
We proposed that dACC, AI, the adjacent caudo-lateral OFC, D2 loop of ventral striatum and LHb form a functional adversity-processing circuit, directed towards inhibitory avoidance and self-control. It learns what is bad or harmful for us, appraises and predicts risks - to stop us choosing, going for and moving for the suboptimal choices which decrease our well-being and survival chances.
Proposed role of dACC is to generate a WARNING signal when things are going (or might end) bad or wrong - leading to negative consequences, to prevent pain, harm, loss or failure. This top-down warning signal from dACC prods for update in our predictions decisions plans (via RLPFC), it enforces the avoidance (D2 loop), corrects (preSMA) or stops (STN) the wrong motor responses, commands for cautious performance, focused attention (FEF, IPS), alertness and alarm (NA, ACh, sympathetic control). The AI signals about bad, low, noxious and aversive biological and social qualities – those which can make us sick or lower our well-being. Cortical adversity-processing regions activate LHb – by direct input and presumably via D2 loop of VS which via GPe, disinhibits GPi - the main source of LHb activation. As the LHb reciprocally inhibits DA release in VTA/SNc and 5-HT in DRN, it stop us choosing and doing things leading to bad consequences but also make us feel down - when overstimulated. By causing low DA and 5-HT levels; and by being attenuated by DA, 5-HT and opioids, this adversity-processing circuit competes with reward-processing circuit for choice behaviour and affective states.
We proposed stimulating effect of DA on active/reactive avoidance and expression of fear anger and violence via amygdala and BNST. Plus calming role of 5-HT and opioids on output of this active avoidance circuit, formed by AI/dACC input to amygdala, and its BNST, PAG, PVN, VTA effectors.
Selected abbreviations and acronyms
AI anterior insula mOFC medial orbitofrontal cortex
APC adversity-processing circuit NA noradrenaline
BNST bed nucleus of stria terminalis NAc nucleus accumbens
CeA central nucleus of amygdala PAG periaqueductal grey
clOFC caudo-lateral orbitofrontal cortex PFC prefrontal cortex
DA dopamine PVN paraventricular nucleus of hypothalamus
dACC dorsal anteriro cingulate cortex RLPFC rostrolateral prefrontal cortex
DS dorsal striatum RMTN rostromedial tegmental nucleus
DRN dorsal raphe nucleus RPC reward-processing circuit
GPe external globus pallidus SNc substantia nigra pars compacta
GPi internal globus pallidus vACC ventral anterior cingulate cortex
LC locus coeruleus VS ventral striatum
LHb lateral habenula VTA ventrotegmental area
MDT mediodorsal thalamus 5-HT serotonin
Adversity processing neural circuit - directed to self-control,
inhibitory avoidance and deselection of the harmful and suboptimal choices.
We characterized properties of the adversity-processing circuit (APC), causally involved in value-based decision making, affective control of choice behaviour, few mental disorders and well-being. We explained why is this APC in competition with the reward-processing circuit (RPC) in guiding goal-directed behaviour. Because APC directly and indirectly attenuates and competes with RPC - via its reciprocal connections, ventral striatum projections and suppression of dopamine (VTA) and serotonin (DRN) release. APC activation causes inhibitory avoidance and deselection of bad, harmful or suboptimal choices, by suppressing the reward circuit output and dopamine signaling. Biasing our decisions, thoughts, actions, emotions.
Proposed circuit includes dACC, anterior insula, adjacent caudolateral OFC, D2 loop of ventral striatum, GPe, subtalamic nucleus, ventral GPi and lateral habenula. Both dACC and AI are reciprocally connected, form part of pain pathway at its top – evaluative processing level, and are commonly co-activated in the NoGo/Go task contrast (inhibitory control task where the prepotent response must get inhibited - withdrawn at the occurrence of infrequent cue).
Figure 1.
The competition of 2 affective
circuits in choice behaviour.
'Go for it' versus 'Stop yourself'
implicit bias – inclination learning
in the motivational VS.
Reward-processing circuit is green.
Adversity-processing circuit is red.
D1 and D2 neurons are spatially
intermixed in the VS.
D2 loop of VS, biasing choice toward
inhibitory avoidance is pink.
Prefrontal projections of mediodorsal
thalamus strengthen the representation
of linked choices and goals.
Dorsal anterior cingulate cortex.
Dorsal ACC is involved in both willful and unconscious control of goal-directed behavior. Many studies demonstrated dACC role in the internal monitoring of actions and performance (Botvinick et al. 2004) and error detection (Carter et al. 1998) – implying it detects conditions under which errors are likely to occur rather than errors themselves. It is also recruited in pain-related and attention demanding tasks (Davis et al. 1997), during response conflict and inhibition of prepotent responses (Barch et al. 2001). Other studies showed dACC role in the regulation of sympathetic system, heart rate and autonomic response to physical and mental stress (Wiliams et al. 2005, Kimmerly et al. 2005, Critchley 2005). Devinsky et al. (1995) divided ACC – depending on its connectivity and control of goal-directed behaviour, into cognitive dorsal ACC region (BA 24', 32') and affective ventral ACC region (BA 25, 24, 33). dACC he found engaged in premotor response selection, in response to noxious stimuli, in cognitively demanding information processing, motivation and sympathetic modulation, and he linked vACC with parasympathetic activation and rest.
Combining the available evidences, dACC seems to identify danger, potentially risky circumstances - and bad or suboptimal outcomes/consequences of our actions, decisions and predictions.
We propose that primal computational role of dACC is to generate a WARNING (Beware!) signal - toward real or predicted bad outcomes/consequences – so to call for changes in current course of our actions and behaviour - to avoid harm or loss. Plus prod for re-thinking and update of the wrong/faulty predictions, guesses, strategies or cognitive models.
The warning signal in dACC is issued whenever things are going bad for us → leading to negative consequences such as pain, harm, loss or failure - to increase our precaution, engagement with the problem, attentional focus and cognitive-motor mobilization. So to improve the performance and prevent the occurrence of bad or suboptimal outcomes - by adjusting our behaviour.
Therefore the purpose of dACC is to learn about and predict bad outcomes and consequences, to warn us about the risks of incurring harms and failures (such as errors in our actions or reasoning processes, thoughts, plans and decisions); so to prevent bodily harm and personal loss. And to urge us to update/adjust the wrong or suboptimal behaviour, strategies and plans, when not appropriate. Thus dACC detects danger and potential risk to prevent bad consequences and harm incurred by current actions, decisions and by behaviour yet to be conducted.
This warning signal generated by dACC resembles negative surprise or unexpected uncertainty signal described by Yu & Dayan (2005) which has been proposed to drive noradrenaline release. Although proposed warning signal is generated also in response to rewarding/attractive stimuli – if there is a need to react to them quickly – to not miss the passing chance (i.e. pretty woman in the street).
Dorsal ACC has all the needed connectivity - not only to identify the risky options and inadequate courses of actions to learn to avoid them - but also to react to potential danger, by recruiting the alarm response of brain and body.
dACC passes its warning signal via top-down projections to several effector regions (Fig. 2). To pre-SMA - to make a fast switch of current motor action, to STN to stop inaccurate motor response. To FEF and PPC to focus our visual and spatial attention. To noradrenergic and cholinergic nuclei to speed up reactivity and increase alertness, to sympathetic system to increase heart and breath rate in demanding situations. To VS, DS, LHb and MDT to bias choice behaviour. It also projects to stress-related hypothalamic nucleus and to threat processing amygdala. By its robust reciprocal connections with DLPFC and RLPFC - dACC warning signal provokes an update/change in failed predictions, decisions or strategies - to switch away from wrong/inadequate behaviour to more optimal choices, actions or performance. RLPFC seeks and tries to find out more fitting model or prediction of what is going on, that can be applied to reach our goals. It generates novel guesses, hypotheses, ideas about studied systems by figuring out relevant relations, extracting contingencies, invariances and rules (IF - THEN) and explaining incongruities in the facts. When testing the validity of new predictions and monitoring their faults (negative feedbacks) - dACC and DS is involved.
Figure 2.
The dACC projections induce
the alarm state, increased
attention and cognitive-motor
mobilization toward its
WARNING signal.
This signal is similar to 'Beware!'
command, evoked by demanding
or risky situations.
The more something matters to us (the more we value it) – the stronger warning signal will dACC issue – when there is potential danger to lose it or fail (i.e. loss of resources, loved ones, social bonds, health, job, prospects, time). So the strength of dACC recruitment will depend on our subjective values and preferences. The warning signal, caused by dACC activation, is possible source of the error-related negativity (ERN). The ERN, an event-related potential, was linked negative feedback and errors (Gehring et al., 1993).
One example for the warning function of dACC is the study of Jurgens et al. (1992). He found that dACC stimulation elicits emotional vocalization in monkeys and humans, while its lesions affect voluntary control of emotional intonation in humans. So this warning signal possibly projects directly to the alarm announcing vocalization circuits, to swiftly warn the conspecifics of impending threat. Another evidence for warning role of dACC is its activation during subjective experience of social rejection (Eisenberger et al., 2003), and impaired avoidance learning of harmful stimuli in dACC lesioned animals (Gabriel et al. 1991).
Results of majority studies support our claim that dACC detects when things go wrong for us - generating warning signal in cognitive, affective, conscious and unconscious appraisal of risky situations and prospects. Thus dACC is selective to learn and warn about affectively bad choices and consequences (pain, harm, loss, failure) and about cognitively wrong (failing) predictions, actions or behaviour. So dACC is causally involved in self-control, as it identifies and reacts to bad or suboptimal outcomes – thus to any threat which could decrease our survival chances and well-being.
For instance, the Shackman et al (2011) review shows dACC activations in response to pain, negative affect and cognitive control. And Eisenberger (2012) demonstrated its role in shared representation of social and physical pain in brain. She also found correlation between its response and the frustration felt by participants about their errors in stop-signal task (Spunt et al., 2012). So pain, bad outcomes and generally any things which went wrong - and we care - are appraised by dACC, leading to inner warning, alarm responses, worries and negative affect of distress. Further supporting findings come from dACC lesioned patients (Critchley, 2003), which do not increase sympathetic activity toward mental stress. Demanding situations and in danger normally activate sympathetic system via dACC, to increase heart and respiration rate.
When dACC is in overdrive i.e. by existential chronic worries or by dysfunctionally low serotonin, opioid or dopamine levels, the repeatedly generated warning, if not attenuated by vACC output - can cause anxiety, lowering baseline DA and 5-HT. Because of its warning role toward potentially harmful or suboptimal outcomes, dACC responds strongly in conflicting, costly, ambiguous and uncertain tasks - as they are prone to lead to errors, loss or harm. They involve higher risks of selecting incorrect or inadequate option or using wrong cognitive model of world to act upon – if solving insufficiently defined task or guessing the right response from incomplete information.
Anterior insula.
Multiple studies showed anterior insula activation in pain (Kwan et al. 2000), discomfort, disgust (Jabbi et al. 2008), aversion (Simmons et al. 2004), empathy with pain (Singer et al., 2006), craving for drugs (Vorrei et al., 2007) and in appraisal tasks. AI receives input from the visual 'What' pathway and from all sensory modalities which bring information about identities of things, their attributes and meanings. AI primarily identifies noxious or unpleasant stimuli linked to pain, possible contamination, sickness or discomfort. AI projects to pre-SMA and premotor cortex, to inhibit/stop response to harmful stimuli. It possibly induces gastric motility in nausea. AI projects to the cognitive rostral and lateral PFC, VS, amygdala and OFC (Craig, 2009). And reciprocally receives and influences the conceptual information, identities and meanings coded in temporal poles.
We propose that AI primarily detects what is BAD or WRONG - harmful, unpleasant, inferior - in the QUALITY of things (+ persons, deeds, social conduct) to avoid those things, which make us feel unwell or sick.
AI learns about biologically noxious stimuli - which cause pain, contamination, discomfort or bodily harm, leading to avoidance or aversion. It also assesses and reacts with rejection to bad, low, degraded or corrupted qualities/attributes of things including aesthetic, social, moral wrongness. Example of such attributes/categories: odd, misfit, ugly, false, suspicious, inconsistent, inappropriate, strange, sick, incongruous, corrupted, faulty, disgusting, repulsive, disharmonious. The unpleasant or noxious stimuli – the approach of which might affect our well-being and health, evoke negative affect – feelings of discomfort which serves to learn to avoid them. Over-stimulation of AI possibly lowers pain threshold, leading to feeling unwell, even to nausea - evoked by strongly aversive stimuli but also by sickening news (rejection or death of loved ones, depravities). The magnitude of pain unpleasantness correlates with AI and dACC activation (Shreckenberger et al. 2005). The phylogeneticaly oldest, innate function of AI, was probably to make us to avoid spoiled food, contagion and contact with infected, genetically or disease-distorted bodies, linked to our biological pursuit of health and well-being and aversion to things which might harm or lower our well-being and health, thus lowering our and our progeny's survival chances. To this primal function linked to aversion and feeling sick was added also social, sexual and cognitive appraisal and judgment. By associative learning, AI learns to represent and react to socially and culturally unacceptable things, deeds and conduct. AI detects also qualitative flaws ans inconsistencies in data - what is odd, not fitting i.e. false statements. This appraisal function is evolutionary linked to the avoidance of potentially bad, harmful things, depending on their biological, social or subjective meanings to us – to minimize their exposure/harm. AI is active in social reasoning tasks, assessing the bad or low quality, aptitude or adequacy of social conduct - targeting what should be avoided or rejected. For example in social cognition/evaluation tasks - when making judgment about outliers, outcasts, injustice (corrupted justice), unfairness or depravity, in moral judgment on what is considered wrong to do to people, when looking down on someone or rejecting him.