The Insular Cortex and the Re-representation of Risk

The Insular Cortex and the Re-representation of Risk

Pauline Meunier PS’23 and Jim Stellar

 

Introduction:

As we have pointed out in other shorter blogs, the brain re-represents its earlier evolutionary mechanisms and functions. What does that mean? It means that as the brain evolved and as it developed it seems to start with the bottom or older structures first. Then higher structures, like the frontal and insular cortex come in later. Often these higher structures do a similar function and the idea is that they re-represent that function and probably control the lower systems to achieve that higher functionality. For example, consider a lower-level superior colliculus brainstem mechanism that directs simple orientation eye movements is integrated with high-level neocortical mechanisms, including likely the frontal eye fields, to consciously direct visual attention. The idea here is that an eyes-on-target brainstem mechanism that might be able to direct a predatory movement strike or an escape reaction in a lizard, becomes in a mammal an object’s visual trajectory. Consider a very young child under 5 months who sees a ball roll behind a screen – they most likely won’t move their eyes away from where the ball disappeared until it comes out the other side. However, a slightly older child moves their eyes to the other side of the screen and waits for the ball to reappear, showing surprise if it does not come out at the right time. We infer that this child has a concept of trajectory, which is a kind of object permanence familiar to developmental psychologists. That such evolutionary older brain mechanisms are still there in adulthood is shown by another example of the infant suckling reflex that reappears in the patients of Denny-Brown (and others) after cortical damage. Much to their embarrassment, these patients will turn and suck the doctor’s finger if the doctor strokes their cheek. The old sucking reflex is still in there.

The question for us here in a blog series on experiential education is how does this re-representation apply to emotional processes related to the experience of an internship. How are they turned into cognitive processes that can confirm or alter a student’s career plans? As an example, how might a negative emotional property of an internship, perhaps mediated by the amygdala, turn into a neocortical representation of a risk of that field of work, perhaps by the insula cortex that can influence a college student’s plans for their field of study? Think of a student who does an internship in a hospital and switched out of the premedical track in college.

As a personal note, in his 2017 book on experiential education, Stellar set forth the insula cortex as a risk detector and put it in theoretical opposition to the non-cortical nucleus accumbens as a reward processor. But this pairing puts a cortical structure in an oppositional balance with a non-cortical structure. And it ignores the issue of re-representation, even as it does point to the opposition of a general approaching and withdrawing neural balance that goes back to an old 1979 paper from his doctoral thesis.

Finally, in this introduction, we again quote the 1600s mathematician and philosopher, Blaze Pascal, that we often use in this blog series, “The heart has reasons of which reason does not know.” The quote supports our general focus on the personal growth in students from experiential education activities in college and the underlying neuroscience of cortical/cognitive interaction with the limbic/emotional system.

 

What is the anatomy of the insula cortex?

The insula cortex, or “Island of Reil,” is a 6-layer neocortical structure buried anatomically within the lateral sulcus (as shown in the figure to the left), hidden away from postmortem view by being folded into the temporal lobe below and the parietal and frontal lobes above. It has connections to emotional, motor, and body sense input and plays a critical function in our evaluative thinking and decision making. The figure below shows two main regions taken from a 2017 review: In red/pink is the Anterior Insular Cortex (AI);  in blue is the Posterior Insular Cortex (PI).

[Note: Abbreviations used in this blog post are listed in a table at the end of the document.]

We note that this review calls the insula cortex an “underestimated brain area” in neuroscience. Also, the precise divisions of the insula cortex are complex and these division continue to emerge as more human brain-scan data are related to anatomical studies from animals and non-human primates, where anatomical tracing post-mortem studies can be conducted.

For example, others in a 2017 review also show the insular cortex, as in the figure above, as being divided into the basic anterior and posterior basic subdivisions and emphasize the separation by a central insular sulcus (CIS). Anatomically, the anterior portion of the insula cortex is proposed to be further comprised of still smaller regions: the anterior short, middle short, and posterior short insular gyri. The posterior portion of the insula cortex is proposed to consist of the posterior and anterior long gyri.

Additionally, the AI (anterior) is also referred to as agranular by cell type, and that is distinct from the posterior (granular) insula cortex, and this review cites an 2010 review to mention a third area – a dorsal posterior dorsal (agranular) insula.

This granular vs agranular distinction is due to the presence of smaller granular nerve cells in layers 2 and 4 of the 6-layer cortical columnar organization. This difference is shown in the figure on the right. In general, agranular cortex is also thought to be older (sometimes called allocortex) and occupies about 10% of the brain whereas the granular (or neocortex) occupies 90%. From the figure, you can see the granular cortex consists of smaller neurons, whereas the agranular cortex consists of relatively larger neurons.

Such a distinction on the basis of the thickness and structure of cortical layers follows an old brain-wide analysis of cortical regions from the early 1900s by Brodman, which has proven useful for functional analysis. Particularly striking in this Brodman analysis is the way the thick input layer 4 in the primary visual cortex abruptly changes density when the visual input from the eyes, via the thalamus, stops entering the cortex indicating the end of visual function. Here in the insular cortex we note that these regional anatomical divisions are the subject of much interest these days in humans, but still rely on work done in monkeys and other animals.

Finally, we would like to point out that a recent study in rats (not in humans) ties the AI to the default mode network (DMN), about which we have written previous blogs. Clearly much work remains to be done on this important area.

 

Insula cortex – inputs and outputs

In terms of inputs, the PI has subregions receiving afferent sensory inputs from both the spinal cord and the brainstem via thalamic networks, as well as inputs from the parietal, occipital, and temporal association cortices, indicating that the PI plays roles in the integration of somatosensory, vestibular, and motor signals.

The AI cortex, on the other hand, forms reciprocal connections to the limbic structures, leading it to be dubbed the “limbic sensory area”. These include notably the amygdala and the cortical regions of the anterior cingulate cortex (ACC) and the dorsolateral prefrontal cortex (DLPFC). We will pay particular attention to this part of the insula given the focus in this blog post on re-representation.

In terms of outputs, the insula cortex has broad connections with the striatum, a vital structure in the organization of movement. This clearly could be important in generating actions, but we also note the close functional connection between emotion and motion itself – even to the point that the word “emotion” contains in it the word “motion.” The insula further innervates the claustrum, a structure known to project information across the whole of the cortex, as well as the hindbrain, which is responsible for the control and maintenance of vital bodily functions, such as heart rate. Innervation of the claustrum may allow the insula to project information regarding our homeostatic needs, or overall internal milieu. Furthermore, innervation of the hindbrain may be what allows the insula to communicate when we are not in homeostasis, triggering automatic functions, as well as homeostatically relevant drives.

As we have spoken about the insula in terms of inputs and outputs, we’d like to make particular note of the connectivity between the insular and the amygdala, striatum, and the thalamus. While this study was done in mice and is therefore not fully reflective of human insular connectivity, the findings are reflective of the function and structure seen in non-human primates and humans.

In terms of insular-amygdala connectivity, the amygdala forms both inhibitory and excitatory bidirectional connections with the insula. Something to note is that afferent connectivity from the amygdala to the insular regions is done via cortex-like structures, rather than striate-structures. These include the basolateral amygdala, the amygdala piriform transition area, the cortical amygdala, and the extended amygdala.

In terms of insular-striatum connectivity, projections of the insula forms efferent connections with the ventral and ventrolateral striatum with converging projections from the piriform cortex, the medial prefrontal cortex, the perirhinal cortex, and the basolateral amygdala. A vast majority of these connections originate from the AI, rather than from the global insula. Furthermore, the AI also innervates the nucleus accumbens core and the interstitial nucleus of the posterior limb of the anterior commissure.

In terms of insular-thalamic connectivity, there are connections observed between the AI, PI, and mPI. Starting with the AI, connections originate primarily from higher-order association and motor nuclei, with the majority of inputs arising from the polymodal association group of thalamic nuclei, including the medio-dorsal and centro-median nucleus. The mPI receives innervation from two sensory-motor related nuclei, including the ventro-medial and ventral anterior-lateral nuclei. Finally, connections to the PI originate primarily from sensory-related nuclei, with the greatest amount of input coming from the posterior complex and the ventral posterior complex. It is important to note that insular outputs reciprocate thalamic inputs, indicating that the insula and thalamus are in constant communication with one another.

 

Some constituent neurons

Pyramidal neurons: The insula appears to be involved in the integration of autonomic and visceral sensory information into emotional, cognitive, and motivational functions. Within both the anterior and posterior insular cortices, there exists a high density of pyramidal neurons. Pyramidal neurons are often known as “projection neurons”, with their axons being sent across the central nervous system. The presence of pyramidal neurons within the insula supports the idea that the insula communicates interoceptive and homeostatically relevant information across the body, as well as integrates various signals from a vast number of cortical regions.

Von Economo neurons: We make a special note that within the AI (and other related brain areas), there is an increased density of what are known as von Economo neurons, a subclass of bipolar neurons, whose function within the insula that were discovered over 90 years ago by Constantin von Economo, but were studied recently by the Altman lab. They are still an ongoing area of research interest as they seem to exist in the insula cortex and other areas, are uniquely found in humans, as well as other higher primates and in dolphins, and have wide-ranging connections in the brain. Antonio Damasio, who wrote about consciousness, thought they were a “divergent-convergent network” that might be involved in integrating emotional and other cognitive representations that could even lead to consciousness itself.

 

Insula cortex anatomical complexity

There continues to be debate about how many structural subdivisions of the insula are present, with a range from two to a maximum of thirteen distinct regions. For example, some suggest thirteen distinctive subdivisions of the insula, including area 52, ParaInsular cortex, the Insula granular, the posterior insular areas PoI1 and PoI2, the frontal opercular areas FOP2, FOP3, FOP4, and FOP5, the middle insular area, the anterior ventral insular area, the anterior agranular insular complex, and the piriform cortex (Pir). In a structure-function review, done within macaque monkeys, 15 regions were identified, whereas in humans, the number ranges from 2-4 primary structural divisions and anywhere from 7-32 further functional subdivisions, as discussed above. We will not discuss these subdivisions, but it is something of note to keep in mind when considering the insula’s structure. Studies done with brain stimulation in humans have identified further structural connections where connections were found from the posterior dorsal insula to the somatosensory areas (motor, somatosensory, and parietal cortices), and from the anterior dorsal insula to several cognitive-emotional areas (e.g., the hippocampus, temporal pole, frontal operculum, orbitofrontal cortex).

Someday, we hope, a true “geographic” network understanding will explain the functional processing of the insula cortex together with other brain regions. Referring to geography, one of us remembers moving to the borough of Queens in New York City and realizing that seemingly every few blocks there was a different ethnic and immigration neighborhood and marveling at how naive he once was to refer to the borough by one name. Maybe this will be our field as it grows less naive.

 

What is the function of the insula cortex?

 This section is naturally even more complex than the previous one on anatomy. As such we will try to restrict ourselves to using human functional data as some reviews point to much variability in the insula cortex in mammals. Also other reviews suggest that humans present a unique opportunity for study.

Human data on insula cortex function comes from three basic sources: 1) activity brain scans like an fMRI, 2)  the behavior of patients with specific insula lesions, and 3) in patients where stimulation is used prior to surgery. In the case of stimulation, given that synchronous activation from an electrode is not the cortex’s natural firing pattern, stimulation is likely to be disruptive (perhaps even like a temporary lesion).

We will focus below on stimulation data and FMRI below, but note here that the exact testing conditions may matter, especially with higher level processing and particularly in humans. For example, the insula in humans is sometimes implicated in empathy, but is it social empathy such as comforting someone on the loss of a family member, or is it action empathy such as helping someone who is physically hurt by springing into action?

A functional review has explored both resting-state and task-related connectivity, as shown in the figure below. They also refer to anatomical track-tracing studies in monkeys (left side) that could not be done in humans. They also note brain scan connectivity studies using diffusion tensor imaging brain scan methods on the right side of the figure.

Comparative trace studies on insular anatomy in humans (middle, right) and rhesus monkeys (left)

 

Three resting-state connectivity functional subdivisions were identified in this review:

1. The posterior insula is functionally connected to the primary and secondary motor and somatosensory areas, especially the secondary and adjacent primary somatosensory cortices.
2. The ventral portion of the AI is further functionally connected to the pregenual ACC, as well as regions of the default mode network (DMN), including the medial prefrontal cortex, the posterior cingulate cortex, and the bilateral angular gyrus.
3. The dorsal AI is connected to the dorsal anterior cingulate cortex (dACC).

Four task-related connectivity subdivisions were identified in this review:

1. A ventral-anterior socio-emotional region, which is responsible for things such as emotional experience, and empathy
2. A dorsal-anterior cognitive region, which plays roles in attention, language, speech, working memory, and other memory tasks
3. A central olfactory-gustatory region, which plays a role in the senses of olfaction and gustation
4. A mid-posterior sensorimotor region,which plays a role in interoception, pain, somatosensory functions, as well as motor functions

In summary: When at rest, the insula activates direct connections with regions of the DMN and SN, networks that are both most active when at rest, as well as regions involved in somatosensory functions. These findings support the idea that the insula is constantly communicating information regarding the body’s internal milieu as well as the body’s periphery. On the flip-side, when we are actively involved in some task, the insula activates connections with a vast majority of the cortex, indicating that when we perform a task, the insula is constantly communicating socio-emotional, cognitive, olfactory, gustatory, and sensorimotor information in order to monitor the body and the task at hand.

 

Looking further at the four task-related functional divisions

The ventral-anterior portion of the Insula (socio-emotional processing) is implicated in socio-emotional processing, especially in how we interpret emotional experiences, empathy, social cognition, and decision-making.

It plays a critical role in how we experience emotions and the subjective feelings of arousal and valence that result from these experiences. Furthermore, the right AI is activated for instances that utilize affective-perceptual empathy, a type of empathy that combines cognitive empathy with emotional cognition. In other words, empathy in which an individual both models and adopts another person’s emotions (via affective sharing). On the other hand, the left AI is activated in instances utilizing affective-perceptual and/or cognitive-affective empathy, a type of empathy that utilizes the theory of mind and refers to the ability to recognize and understand another person’s mental state.

Both the right and left AI are activated in response to seeing others in pain and in response to others’ expressions of disgust, fear, anxiety, happiness, and sadness, all of which are generally considered basic emotions. Finally, the vAI has been implicated in the process of making risky decisions, especially in terms of the emotional aspects of decisions, which will be touched on in applications to experiential learning.

The dorsal-anterior insular cortex (dAI) (cognitive – salience and attention) has been most often implicated in a number of cognitive processes, including, but not limited to, attentional processing, salience processing, and speech. The dAI is also thought to be involved in the motor planning of speech, as shown in patients suffering from isolated insula lesions following a stroke. There may also be insular involvement in the production of speech, but it is not yet known the extent of it’s role.

The dAI facilitates the detection of novel stimuli within the environment, through its role in the “salience network (SN)” alongside the dACC, amygdala, hypothalamus, ventral striatum, thalamus, and specific brainstem nuclei. The SN functions to identify the most homeostatically relevant stimuli among multiple internal/external signals, and we will be delving into much further detail later into this blog.

Furthermore, the dAI has influences over structures and activities of the DMN that is responsible for self-relevant and social cognitive processes. It also has influence over the central-executive network (CEN) that is responsible for the maintenance and manipulation of afferent signals in decision-making.

The Central Insular function – (olfactory-gustatory region) (CI) is also known as the gustatory cortex, as well as believed to play a major contributing role in the actions of the somatosensory cortex in humans, as well as rodents and non-human primates. The CI has been implicated in playing a major role in the integration and re-representation of chemosensory stimuli, especially olfactory and gustatory.

Firstly, in both rodents and humans, the CI is thought to be the brain’s possible flavor center, with relay neurons from the ventral posteromedial thalamic nucleus (a major structure in the integration of both gustatory and olfactory stimuli) projecting their axons to the intermediate dysgranular as well as the granular insular cortex (mentioned previously). In rodents, there are bidirectional pathways between the insular cortex and the Pir (an intermediate structure in the gustatory pathway), as well as anterograde projections from the posterior AI reaching the intermediate dysgranular insular cortex as well as the Pir. Studies that performed bilateral lesions of rodent gustatory cortices found that lesioning led to a reduction in taste memory reflected through a reduction in the ability to adequate modulate the incentive reward value of a given food stimuli. Further neuroimaging studies of rodents, non-human primates, and humans have found overlapping activation of the CI in response to sweet smells and tastes, suggesting that there is olfactory representation of tastant stimuli within this region of the insula. Furthermore, the sweetness intensity of a given stimuli is directly positively correlated with the degree of insular activation seen, indicating that the insula may be selectively sensitive to sweet tastes/odors.

In terms of olfactory processes, research has demonstrated that insular gray matter volume is directly related to an individual’s olfactory function, with lesioning studies demonstrating that isolated insula lesions led to impairments in taste and odor-induced taste perception. In humans, neuroimaging has found a specific region around the central insular sulcus that may be responsible for the processing of unpleasant odors.

The mid-posterior insula (mPI)  (sensorimotor integration) plays roles in sensorimotor processes, such as visceral sensations, autonomic processing, interoception, somatic processing, pain processing, auditory processing, chemosensory functions, and vestibular processing. The mPI is widely considered a central visceral-sensorimotor region, receiving afferent projections and transmitting interoceptive signals from across the entire body. The transmission of interoceptive signals to and from the insula cortex is believed to play a large role in the maintenance of homeostasis and even autonomic bodily functions that are mediated by lower brainstem areas. The most commonly accepted global function of the insula revolves around interoceptive monitoring of the body’s homeostatic functions. Interoception is defined as the collection of senses that communicate information about an organism’s internal state, with input from all major organs. Interoception functions to keep our bodies working towards survival, and includes functions such as monitoring hunger and fullness cues, heart rate, breathing rate, etc. It may be the reason when risk is undertaken there is a “butterflies in the stomach” sensation.

The mPI also works in processing somatosensory information from the body’s external periphery, prompting somatosensory manifestations within the body, most predominately in the contralateral face and arm regions, with some influence over ipsilateral, bilateral, and midline structures. Further research has demonstrated that the dorsal margin of the contralateral middle and parietal insula is activated differentially for both non-painful tactile stimuli and painful tactile stimuli, without the involvement of the parietal somatosensory structures.

Auditory stimuli are also processed within the mPI, with efferent connections to the primary auditory, auditory association, and post-auditory cortices. It has been demonstrated that direct stimulation of the ventral posterior insula (vPI) leads to auditory illusions and distortions, such as hyperacusis (increased noise sensitivity). Furthermore, isolated insular damage has shown to lead to unilateral deficits in auditory speech processing, as well as the development of non-verbal auditory agnosias.Within this region, the intensity, quality, and effective value of taste stimuli are processed. Furthermore, the PI is activated in response to taste along with other areas, e.g the piriform and orbitofrontal cortex, the amygdala, and the ventral putamen. Finally, the PI is believed to make up a vestibular region along with the parieto-insular vestibular cortex (PIVC) together known as the PIVC+. This PIVC+ region is believed to be a core region of the vestibular cortex, and therefore a key player in vestibular processing, located in the mid-posterior Sylvian fissure. The PIVC+ receives visual and vestibular signals from a variety of subcortical structures, most notably the lateral thalamus nuclei (especially the ventral-posterior nuclei), and the pulvinar. Much of the PIVC and PI involvement in vestibular processing is a matter of speculation. However, some researchers believe the PIVC encodes head and full-body movement, as well as estimates directional movement by means of those encoded signals. The PI, on the other hand, is believed to be involved in the estimation of directional movement via the combination of visual and vestibular cues, as well as the distinction between visual self-movement and visual object movement.

 

In general

In terms of general Insula cortex reviews, a review in 2015, suggests that in addition to the prefrontal and parietal cortices, several subregions within the insula, including the PIC, vAI, and dAI, are activated during the process of decision-making. We see in this same review, the authors present what they call a temporal map of decision-making with four stages: 1) refocusing attention, 2) evaluation, 3) action, and 4) outcome processing. A 2005 review suggests a similar three set of stages but really beginning with evaluation which they call formation of preferences (or step 2 above). The convergence here is encouraging. Both reviews emphasize somatic affective components of making this decision.

The AI is said to be involved in the affective component of the first stages of refocusing and forming an evaluation or a preference. In doing that, it interacts with the amygdala to generate changes in autonomic function and with the orbital frontal cortex to generate subjective feelings and what they say is interoceptive awareness. In stage 3, experiencing the outcomes, the insula is said to be involved with feedback, again interacting with the amygdala, accumbens, orbitofrontal cortex, and particularly the medial prefrontal cortex. Their review does not have a role for the insula cortex in stage 2.

As automatic affective appraisals are made, an additional intermediate step elicits the generation of somatic markers, which signal the intensity (saliency) and valence (positive/negative) of a given stimulus. While the overarching impact these somatic markers have on present/future decision making are still debated, it is believed they allow for the encoding and recall of previously experienced salient sensory stimuli through the process of long-term potentiation. The insula cortex, in concert with the amygdala, elicits the generation of somatic markers, contextualized by observed changes in the autonomic nervous system (i.e., changes in skin conductance, heart rate, blood pressure, etc.). It has also been demonstrated that somatic markers send feedback signals back to the cortical structures, especially the insular and orbitofrontal cortices. These somatic markers participate in decisions, and although this study looked at patients with more common ventromedial prefrontal cortex damage, not insula cortex damage, their diagrams involve the insula cortex as giving rise to “conscious gut feelings.”

The involvement of the amygdala with its classic role in fear and fear learning suggests that the interactions with the insula cortex are possibly where the negative emotional generated by the amygdala turns into something more symbolic, i.e. potential risk to be balanced against reward, in the cortical network in which the insula cortex participates. Since loss avoidance has long been seen in economic decision making, it could be the role of the insula cortex to guide decision making from this perspective.

                                                     

Back to von Economo neurons and their potential function

The pathways via which the salience network conveys information to the rest of the brain are unique in that they are the only location of the human cortex to contain a high concentration of a specialized subclass of bipolar neurons, von Economo neurons. They are large, bipolar neurons located in layer 5 of the ACC and the AI, with significantly fewer spines, intersections, and shorter dendritic length compared to other layer 5 pyramidal neurons, indicating that von Economo neurons may receive and integrate less information compared to pyramidal neurons. However, cell bodies of von Economo neurons are approximately 4.6 times larger than those of pyramidal neurons, as well as have an abundance of non-phosphorylated neurofilaments, both of which are characteristic of neurons with large axonal surface area. These von Economo neurons have been shown to project their axons into white matter tracts, which when combined with a large axonal surface area, suggests that their function may be to rapidly relay signals processed in the ACC and the AI across the brain. The figure at the right shows a comparison from a study of vonEconomo neurons and modified pyramidal neurons in the fronto-insular, ACC, and prefrontal cortex with Golgi and Nissl staining

Re-representation, risk, and the insula cortex

To restate, our goal in this blog post is that incorporating risk into explicit cognitive/cortical decision-making involves a re-representation of simpler emotional processes that may come out of basic limbic system emotional processes (e.g. fear learning in the amygdala). This kind of re-representation supports the general cognitive symbolic logic that is necessary for explicit planning such as the choice of or change of a major in college after an experiential activity like an internship. It also supports the acknowledged importance of reflection in experiential education thinking to better integrate cognitive and emotional processing as we have written about in a past blog.

To define risk, in his 1921 book, Risk, Uncertainty, and Profit, economist  F.H. Knight defined risky decisions as those in which “the decider has priori or statistical knowledge about the probabilities of likely outcomes.” Risk-reward balance is an obvious key in making decisions and much has been written about it recently by Kahneman in his best-selling 2011 book, Thinking Fast and Slow. Risk also comes into play in the phenomenon of loss aversion and in Kahneman’s work on heuristics or short-cuts in that process, we see as influence of implicit emotional or limbic processing. His “thinking slow” is the deliberate or explicit cognitive processing that allows the comparison of risks and rewards in the future as according to the symbolic and explicit logical that leads to plans. The integration between thinking fast and slow, we think, depends upon re-representation and the cortical component of that integration we say is in the insula cortex.

 

Risk and insula function in decision-making

In a 2015 review on the insula cortex and general behavioral decision-making, four stages of function are given:

1. re-focusing attention
2. evaluation
3. action
4. outcome processing.

Other authors have three stages that are similar but a bit different:

1. formation of preferences among options
2. selection and execution of an action
3. experience or evaluation of an outcome.

We will follow the four stages here from the 2015 review, but in this post combine the first two into one discussion point. Interestingly, these stages of decision-making are reflective of the stages involved in the experience of an emotion episode, these being (1) eliciting event, (2) physiological/subjective feelings change, (3) cognitive appraisal formation, and (4) behavior change/action.

1) Re-focusing attention and evaluation involve the somatic marker hypothesis that posits that the insula cortex is sensitive to somatic markers that arise in response to changes in the body’s periphery, likely triggered by the limbic system. When evaluating particularly complex stimuli, previously learned and stored somatic marker patterns are triggered and can be useful in avoiding negative outcomes. For example, in a 2016 study, when subjects are trying to decide between options of decks of cards that give and take away money, there is a somatic reaction when the hand is about to touch a deck and that influences the cognitive decision-making of which deck is optimal. Such reactions to stimuli patterns alert the subject to something important going on or can help them select between options as in this task.

When it comes to making risky decision-making, activation of the dAI and vAI has been observed. Activity within the dAI has been demonstrated to be positively correlated to the actual outcome, which may be a result of AI’s role in coordinating and directing attention towards relevant stimuli. Additionally, human work with brain lesioned patients has identified the insula cortex and other frontal cortex areas as working together on various aspects of risk estimation and risk adjustment  with damage to the insula cortex particularly involved in decreasing risky behaviors as risk increases. This seems a particularly fertile field as risk assessment can be studied in monkeys with the complex stimulus expected evaluations involving magnitude, variance, and skewness that lead to future decision-making, where more control over the insula cortex can be achieved. The dAI also plays roles in tracking risk prediction, as well as risk prediction errors in  human EEG studies as well as fMRI studies.

Going back to the 2015 review cited above, when activation was being observed during completion of a gambling paradigm, dAI activation was indicative of choosing a safe option. Anticipation of uncertain positive/negative rewards elicits activation of select regions within the dAI and vAI, with the activation being correlated to self-reported arousal, and in cases of an anticipated negative outcome, predicts negative outcome evaluation.

2) Action selection involves differentiating between intentional and stimulus driven behavior. Intentional actions are distinguished from those stimulus-driven ones in that they are not triggered by change in the body’s periphery, and are instead internally generated and motivated. Intentional action has been demonstrated to be made up of three different components, namely the “when”, “what”, and “whether” components. The “when” component regards the choice of action timing, as influenced by external stimuli. The “what” component involves the selection of the best option among several alternatives. Finally, the “whether” component is deciding whether an action should be taken, and can involve the conscious consideration of whether or not to engage in a particular action. In general, the dAI is involved in the “what”, “where” and “whether” components. Strong bilateral activation of the dAI is present when comparing internally triggered choices with externally triggered ones. Furthermore, when comparing a stimulus-driven versus a free-choice option, activation of the right vAI and left dAI increases. Within the “whether” component, activity of the global AI is thought to be a major player in inhibitory control over action urges, with activation of the vAI signifying the intentional ending of a task.

3) Outcome processing in decision-making is the processing as well as the storing of information for future harm reduction and risk prevention. This is the point where we would say the re-representation of limbic processes for risk is being used with other cortical regions to make plans. The insular cortex has been demonstrated to be a key neural structure underlying error awareness, with consistent AI activation present during the commission of errors as well as in cases with required performance monitoring. Being aware of when we have committed an error in our decision-making is vital to our survival as it triggers post-error corrective behavior. Further research has shown that increased bilateral activation of the dAI is present when comparing conscious and unconscious error commission, suggesting that the AI recruits necessary cortical resources once an error has been accurately detected. Also, error awareness following an incorrect decision-making leads to decreased response time in subsequent trials as a result of increased activation of the dAI, DLPFC, and FPC.

In the context of social decision-making, the insular cortex also has a role in outcome processing, focusing on post-error slowing, harm reduction, and harm avoidance. In studies in which cooperative participants were faced with non-cooperative confederates, unreciprocated cooperation was associated with increased activation of the dAI and vAI within the cooperating partner, expressed through the eliciting of negative emotions towards the non-reciprocating confederate. Furthermore, functional connectivity between the AI and lateral OFC predicts defection in future interactions with that same non-reciprocating individual. Researchers have described the implications for a separate subnetwork composed of the lateral OFC, AI, amygdala, and hippocampus, that activates in response to violations of social-contracts, termed the “Ventral-Anterior Insular Network”. While the existence of this network is still a subject of research, it has significant implications for the resource-recruiting role of the anterior insula cortex similar to those observed as a result of post-error slowing. Further work has demonstrated when comparing fair and unfair choices, there was increased activation in the AI, DLPFC, and ACC, with AI activation scaled to unfairness magnitude and strongly correlated with rejection rate of unfair offers.

In the harm reduction component of outcome processing, the insula plays roles in evaluating potential risk or uncertainty in decision-making in order to maximize homeostatic goals. Making a risky decision elicits increased activation of the dAI during the feedback stage. On the flip-side, if an individual has previously made a risky decision in a situation, the insular cortex mediates the risk decreasing effect that increases the attractiveness of that same risky option. Furthermore, during the decision-making stage, slight deactivation of the dAI is present in individuals who have previously taken a risk, reflected through increased selection of risky options. Finally, increased activation of the vAI bilaterally during non-risky decision-making consistently led to increased risky-decision-making.

 

Summary

Taken all together, activation of the insular cortex fluctuates in a “see-saw” like fashion: Activation increases directly before taking a risk, leading to a plateau in activity if a safe option is chosen. Choosing the safe option elicits the generation of action urges to make the risky decision, thus increasing insula cortex activity and subsequently increasing the likelihood of making a risky-decision.

When an incoming novel salient external stimulus causes a change in the body’s periphery, there is an increase in AI activity. The dACC then selectively amplifies these signals, leading to signal synchronization within the right fronto-insular cortex (rFIC), mediating switching to either the DMN or CEN depending on the nature of the stimulus. For intentional actions stemming from externally-driven stimuli activate the CEN, whereas non-intentional actions stemming from internally driven signals activate the DMN. This dynamic switching brings attentional focus to the stimuli, and starts the decision-making process. The initial increase in AI activity leads to an increase in risk prediction, and assuming there are no previously stored somatic markers, a safe option will be selected, leading to a decrease in PI, dAI, and vAI activity. The subsequent outcome is then processed, including associated subjective feelings and action urges, causing an increase in dAI and vAI activity as the previously discussed “what”, “when”, and “whether” components of action are processed. Regardless of the action’s outcome, the amygdala and hippocampus then store the experience in terms of somatic markers into long-term memory. If a similar stimulus is presented in the future, the increased AI activity can trigger the recall of previously stored somatic markers from long-term memory. Recall leads to a decrease in risk prediction (if the outcome was positive), leading to an increase in urge generation to select a riskier option, and a subsequent increase in PI, dAI, and vAI activity.

As stated above, the general purpose of this blog series is to look at how experiential education in college students informs and alters the classical academic (cognitive) plan of study for a field. With the well-known high impact of experiential activities outside the classroom on the limbic system, the question becomes, how does this process get integrated with the cognitive evaluations of activities and plans that go into one’s chosen major or a change in that major. The hope is that a detailed understanding of the underlying neural processes will not only help to understand the educational dynamics here, but will also help to increase the salience to all about the importance of this behavioral work to what we sometimes call the “brain natural” way of fostering a student’s intellectual and professional growth in college. We hope we achieved that here by going deep into the behavioral neuroscience of the insula cortex and the particular phenomenon of risk and the insula cortex processing of that risk as a re-representation of lower-level more emotionally reactive processes of the limbic system from that experience.