Human behavior is often explained in terms of unseen entities such as motivation, curiosity, anxiety and confidence.
What has been unclear is whether these mental entities are coded by specific neurons in specific areas of the brain.
Professor Adam Kepecs at Cold Spring Harbor Laboratory has answered some of these questions in new research published in Nature. The findings could lead to the development of more effective treatments for obsessive compulsive disorder, compulsive gambling and other psychiatric disorders.
The team studied the orbitofrontal cortex, an area critical for decision-making in humans and animals alike. Damage to this brain region impairs decision-making. In a famous example, Phineas Gage, a railway worker, survived extreme damage to this area when an iron rod pierced his skull in an explosion. Gage survived but his personality and decision-making skills didn’t.
Kepecs and his lab set out to clarify how neurons in orbitofrontal cortex encode mental variables such as motivation or confidence.
“We wanted to understand how neurons code for these mysterious entities, what’s the logic behind it, what’s the architecture of orbitofrontal cortex,” Kepecs said. By monitoring the neuronal activity in the brains of rats making complex decisions, the team identified novel, unsuspected structure in the functional organization of orbitofrontal cortex.
Researchers seek to determine what messages such neural activity represents. The standard approach is to determine which features of the world a neuron cares about (which ones increase their activity) and which features they are insensitive to (no change in activity).
For instance, in the visual cortex, neurons are tuned to the edges of objects and each neuron has a preference for differently oriented edges. “When you’re dealing with mental variables, the question becomes, “How do I define them?'” Kepecs said.
“How do I really know that this is confidence or this is value?”
The key insight was to use mathematical models of choice behavior to compute ‘decision confidence.”
This approach yielded quite specific predictions about what a representation of confidence looks like in terms of observed variables such as the difficulty of the decision or which choice was made.
It turned out that many orbitofrontal neurons were consistent with these predictions, their activity increasing or decreasing with formally defined decision confidence.
Previous studies of the orbitofrontal cortex have identified similar mental variables, but unlike in other brain regions such as the visual cortex, there was no order in their responses and the complexity of coding was baffling. “What people found was a mess,” said Kepecs, who has spent his career understanding neural underpinnings of decision-making.
To make sense of this mess, Kepecs and his team took a different approach than others.
Postdoctoral fellow Junya Hirokawa, now an associate professor at Doshisha University in Kyoto, Japan, recorded large populations of orbitofrontal cortex neurons and used sophisticated machine learning techniques to understand their activity patterns.
The team discovered that neurons fell into distinct functional groups. And each group of neurons coded for different mental variables, like decision confidence or reward value, revealing a hitherto unsuspected highly structured organization.
Finally, Kepecs wondered whether these functional groups are supported by specialized anatomical structure. To do so, the team used engineered viruses to target a specific group of neurons, those that send connections to the striatum, a part of the brain important for updating or rethinking the value of a choice. They monitored the activity of these neurons and found that these coded for another mental variable, reward value, increasing activity when the expected reward was low.
“What that does for us is it says, “Look, there might be an anatomical logic that’s part of this functional logic,'” Kepecs said.
Deconstructing this logical relationship between how the neurons function during different tasks and how they are physically structured in the brain could also open up such possibilities as treatments for psychiatric disorders, such as more accurately stimulating the brains of patients with severe depression, Parkinson’s, and other types of diseases.
Primary insomnia (PI) is a chronic clinical symptom characterized by the subjective experience of sleep loss and disturbed sleep. Patients with PI show heightened arousal and find it difficult to sleep in bed. It is a very common sleep disorder among the general population. Although many people with PI do not have any identifiable psychological or psychiatric problems, there is evidence to suggest that untreated PI may be important in the development of psychiatric illnesses, such as substance abuse and depression.[2,3] Moreover, insomnia can lead to impairments of many basic cognitive functions, including learning and memory, attention,[5,6] as well as emotional impairments.
Over the past 20 years, the relationship between sleep and memory has attracted much attention, and several reports have confirmed that sleep is important for memory processing.[8–10] Taking a nap can improve memory, which supports the view that even short-term sleep is advantageous for memory consolidation. Memory impairment is thought to be the core symptom of the decline in cognitive function associated with sleep loss[12,13] and may encompass deficits in working memory,[14,15] as well as encoding of new memory information. Functional imaging studies have suggested that memory impairment in patients with PI may be related to decreased brain function in the temporal cortex and frontoparietal network.[17–20]
In recent years, many studies on sleep disorders[21–23] have implied that sleep is also important in emotional memory, and that sleep loss negatively influences emotional memory, but not the categorization of emotional perception. Some studies have indicated that emotional information is remembered better than neutral information, and that there may be a preferential consolidation of emotional memory, as compared to neutral memory, during sleep.[26–28] In the past few years, it has been suggested that sleep loss has a greater negative effect on the memory of positive and neutral than of negative stimuli.[29,30]
The formation of emotional memory depends on the activity in specific structures, such as the amygdala, insula, prefrontal cortex, and hippocampus. A study by Motomura et al  showed increased activity in the amygdala in sleep-deprived subjects when they were presented with aversive pictures.
Baglioni et al also found that the reactivity of the amygdala to negative stimuli does not seem to be impaired in patients with insomnia. The above studies indicate that sleep disorders can lead to amygdala reactivity, especially after exposure to negative emotional stimuli.[31,33
] The amygdala is a key brain region in emotion processing, as it not only connects with many other emotion-processing regions, but also integrates local and global networks involved in emotional and cognitive information processing.[31,34]
A study by Shao et al  showed that sleep deprivation affects the emotion-processing circuit and decreases the functional connectivity between the prefrontal cortex or anterior cingulated cortex and the amygdala. The altered functional connectivity between the amygdala and other brain regions may be dedicated to processing of emotional memory with different valences.
Another functional change includes the prefrontal lobe, which is more prone to be affected by sleep loss. A study by Thomas et al  showed that, after 24 hours of sleep deprivation, there is a significant decrease in metabolic activity in the prefrontal cortex, including the orbitofrontal regions, which are involved in decision-making under conditions of uncertainty, such as that required for the Iowa Gambling Task (IGT). Altena et al also revealed that patients with PI exhibit smaller gray matter volumes in the left orbitofrontal region, a finding that strongly correlated with the subjective severity of insomnia.
A wide body of literature has provided evidence of the neural mechanisms underlying IGT performance, which involve the function of the prefrontal cortex, and especially of the orbitofrontal regions.[39–41] Moreover, several studies have reported that this neural circuitry maybe sensitive to insomnia.[19,42–45]
Previous studies have shown impaired decision-making ability in the IGT in participants with sleep deprivation, as evidenced by their increased number of choices from disadvantageous decks.[46,47] A recent study from Seeley et al suggested that sleep improves strategy-decision learning ability in the IGT; these results provide new insights into the relationship between sleep and IGT learning.
Decision-making in the IGT is associated with the orbitofrontal regions,[40,49] and decision-making ability has been shown to be impaired in participants with sleep deprivation.[46,50] Previous studies have also confirmed functional abnormalities in the prefrontal cortex of patients with PI.[38,45]
These functional abnormalities may underlie the significant cognitive deficits associated with PI, which may include deficits in emotional memory and decision-making. However, whether PI shows analogous outcomes in emotional memory and decision-making, similar to sleep deprivation, remains unclear.[21,22,46]
In the current study, we hypothesized that patients with PI would have deficits in emotional memory of different valences and in decision-making. We administered an emotional picture recognition task that included a phase of emotional picture evaluation and a delayed recognition phase. In order to investigate whether emotional memory impairment is attributed to emotional perception, we also asked participants to evaluate the valence and arousal of the emotional pictures with scores during the emotional picture evaluation phase. We also tested the decision-making ability of participants using the IGT and a series of neuropsychological tests, to determine if the above cognitive deficits could be detected in patients with PI.
Discussion and conclusions
The aim of the present study was to investigate the effects of PI on emotional memory and decision-making. As hypothesized, our results indicated a generalized deleterious effect of PI on emotional memory, while decision-making ability was also significantly impaired in patients with PI.
We first assessed the influence of PI on emotional memory. Patients in the PI group performed well in the emotional picture evaluation task, but not in the emotional recognition task. These results are consistent with previous studies reporting impairment in emotional memory in patients with sleep deprivation.[21,22,59]
Other studies have indicated that sleep plays a key role in emotional memory processing.[24,26,27] Cunningham et al reported that sleep can lead to preferential consolidation of negative emotional memory. A study by Cellini et al indicated that daytime napping is beneficial for consolidating emotional memory presented before and after sleep, irrespective of valence. However, poor sleep quality has been reported to affect the emotional valence of memory negatively.
Previous studies on the relationship between sleep and emotional memory have mostly utilized experimental methods testing sleep deprivation[22,60] or have included only HC subjects.[61,62]
Although many studies have shown that emotion or memory[18,64] may change in patients with PI, there has been little research on emotional memory in these patients. Our findings imply that patients with PI have impaired performance on emotional memory tasks.
We also found different patterns of alterations in the recognition of emotional pictures, as a function of their valance. Specifically, we showed that PI adversely affects recognition of positive and neutral pictures, but not of negative pictures.
Previous studies have suggested that negative stimuli may be remembered better than other stimuli.[21,65,66] A study by Tempesta et al indicated that the accuracy of remembering negative pictures is more stable than that of remembering neutral or positive emotional pictures, in subjects with sleep deprivation.
This might be attributed to the facilitating effect of negative stimuli during the encoding phase, which seems to be mediated by the amygdala.
Moreover, previous studies have shown that negative stimuli increase activity in the amygdala during sleep loss.[32,33] This indicates that the greater stability of negative stimulus memory may be attributed to the increased reactivity to negative pictures, induced by insomnia, and the heightened amygdala responses to negative stimuli.
Our results are similar to those of previous studies on individuals with sleep deprivation.[21,29] Previous studies have suggested that emotional memory relies on many brain regions, including the hippocampus, prefrontal cortex, and amygdala,[31,69] and that sleep loss negatively affects the functionality of these regions.
Structural and functional neuroimaging studies have provided insight into the alterations of regional brain function in PI. These studies have shown that gray matter in the orbitofrontal and cingulate cortex, hippocampus, and middle temporal gyri is affected by chronic PI.[38,71,72] Changes due to PI include decreased connectivity in the frontoparietal network[19,43] and emotional circuits.
The deleterious effects of PI on memory retention of positive and neutral pictures, but not of negative pictures, may be related to the amplified reactivity of the amygdala to negative stimuli,[25,74] as well as the decreased functional connectivity of the amygdala with the prefrontal cortex.[35,75,76]
The second aim of the present study was to assess the social decision-making ability in patients with PI. Patients with PI had significantly impaired performance in the IGT. In this task, the PI group more often selected disadvantageous cards and placed higher bets, based on simple probabilistic decisions. In the first block, participants in both the PI and HC groups tended to select more disadvantageous cards.
This result suggests that the 2 groups were unaware of the rule at the beginning. However, analyses showed that the HC group improved significantly by making advantageous choices in the test phases over the PI group.
The results indicated that the participants in the HC group could gradually shift their more disadvantageous card selections toward the advantageous decks after the first learning phase, but the PI group did not show this beneficial behavior pattern and failed to learn the rules to select cards from advantageous decks.
Two studies from Killgore et al[46,77] also showed that, after sleep deprivation, volunteers tended to choose from the disadvantageous high-risk deck more frequently. An event-related potential study showed that the N250–400 amplitude was smaller after sleep deprivation in the feedback stage of the IGT.
The results suggested that sleep loss affects risk-taking behavior due to reduced individual responses to negative feedback stimuli. Both Pace-Schott et al and Seeley et al provided new evidence that sufficient sleep can improve understanding of decision-making rules, as well as behavioral outcomes. Decision-making ability under conditions of uncertainty, as assessed using the IGT task, has been proven to be sensitive to abnormal functioning of the orbitofrontal cortex (OFC) and ventromedial prefrontal cortex.[79–81]
Several functional brain imaging studies have also demonstrated that the medial frontal cortex is activated when subjects perform decision-making tasks under uncertainty.[82–84]
Moreover, patients with dysfunction in the ventromedial prefrontal cortex, including regions of the OFC, fail to develop anticipatory electrodermal responses before making a choice.
This, in turn, disrupts the ability to utilize emotional signals to guide decision-making, to learn from past experiences, and to avoid adverse choices.[85,86] Functional imaging studies have confirmed that PI can lead to structural changes in the prefrontal cortex,[38,71] decreased low-frequency fluctuations in the bilateral OFC, and reduced functional connectivity to emotional circuits.[73,88]
Our data suggested that participants with PI have similar deficits as patients with damage to the OFC.
Although patients with PI show fewer global impairments than do patients with brain injuries, as seen in a clinical setting, they exhibit similar performance patterns as patients with OFC lesions. This suggests that the functioning of similar prefrontal cortical regions may be adversely affected by PI; however, in the present study, we did not provide evidence for a direct reduction in prefrontal activity in the PI group during the IGT.
Apart from deficits in emotional memory and decision-making, we found that patients with PI exhibited widespread basic cognitive impairments, including deficits in working memory, and executive function. These results are in line with those of previous studies.[42,89]
By using Pearson’s correlation analysis, we also showed that sleep efficiency correlated with the results of the trail-making test, consistent with the results of a previous study that indicated that poorer sleep quality is associated with poorer executive function.
Moreover, the number of advantageous choices in the IGT correlated significantly with the BAI score. Previous studies have shown that patients with trait anxiety show a choice preference for deck 1 or 2 in the IGT, suggesting that deficits in social decision-making ability may be due to exaggeration of emotional feelings or emotion regulation deficits.[92,93]
Emotional changes are associated with punishment or reward signals for the past and potential occurrence of an outcome, thus guiding long-term behavior according to the “somatic-marker hypothesis”. In line with previous research reports, our findings suggest that anxiety, caused by PI, can lead to impaired decision-making in terms of risk under ambiguous conditions.
This study had some limitations. In the absence of functional brain imaging data, we could not provide direct evidence to demonstrate whether deficits in emotional memory and decision-making in patients with PI are due to functional changes in the amygdala and prefrontal cortex. The PI and HC patients were not adequately matched at baseline, especially in terms of the psychological state of both groups, although analysis of covariance showed that the BDI and BAI score had no covariate effect on the accuracy (d′ values) for positive, neutral, or negative pictures or on the total number of advantageous choices in the IGT.
Therefore, further neuropsychological and functional brain imaging studies are required to confirm the neural mechanisms underlying emotional memory and decision-making impairments in patients with PI.
Our findings suggest that insomnia had different effects on memory, depending on the emotional valence of the memory. Specifically, there was memory performance impairment for positive and neutral items, but the recognition of negative stimuli seemed to be more resistant to the effects of insomnia.
Our results also suggested that decision-making, including decision-making under conditions of uncertainty, may be vulnerable to PI. However, elucidating the relationship between emotional memory and decision-making impairment in patients with PI and changes in prefrontal lobe function will require further functional brain imaging studies.
More information: Junya Hirokawa et al. Frontal cortex neuron types categorically encode single decision variables, Nature (2019). DOI: 10.1038/s41586-019-1816-9