This is my review of a recent paper from Prof. Rebecca Saxe’s lab at MIT’s Department of Brain and Cognitive Sciences. The authors (Tomova et al. 2020) present a study investigating if forced isolation makes people crave social interactions similar to how a hungry person craves food. This study is extremely relevant in the current scenario, where millions of people worldwide have been forced to self-isolate.

Introduction:

Around two decades back, the conditions of orphanages in Romania were not so good. Dr. Charles Nelson, a professor at Boston Children’s hospital, was horrified upon seeing the conditions on his visit. He reported that every 15 children had a single caretaker, and they were instructed not to pick them up, attend to them, or show affection in any way, even if they cried. They weren’t even played with. This was instructed given the concern that any show of affection will make them want it more, and with limited workers, it would be hard.

As expected, these children developed ‘indiscriminate friendliness.’ This behavior could be because such chronic social isolation must have made them crave emotional care and social interaction. This account raises the importance of social groups and interaction and how their absence affects us.

Studies so far confirm that loneliness is associated with lower physical and mental health. Social interactions may be basic needs for humans, like food or sleep. If that is so, their absence should elicit a ‘craving,’ which motivates behavior to seek what is lacking. Even though it has been shown that the reward system in our brain is activated when cues with positive social interaction are presented, studies on neural correlates of the effect of mandatory social isolation are rare.

When mice, for example, are subjected to social isolation, a ‘loneliness’ brain state develops, and dopaminergic (DA) neurons mediate the behavior to seek social interactions, also called ‘craving’ behavior.

Hypothesis:

The hypothesis being investigated here is whether a similar response (to mice), i.e., social craving analogous to the craving for food, after fasting is evoked, and dopaminergic (DA) neurons mediate the ‘craving’ behavior. This has never been directly tested in humans before.

Methods:

To study this, 40 socially connected young adults underwent three sessions of the experiment:

1. Isolation: 10 hours of social isolation
2. Fasting: 10 hours of food fasting
3. Baseline: Functional localizer task Each of these sessions was followed by an fMRI scan with cue-induced craving (CIC) paradigm. In the CIC, cues for social contact (e.g., Friends sitting together and laughing), food (e.g., pasta), and control (e.g., flowers) were presented. After each block of cues, participants self-reported their ratings for cravings concerning the block presented. Two things that I believe are very important in the context of the study and were taken care of:
4. According to the University of California, Los Angeles (UCLA) loneliness scale, the participants had low levels of pre-existing loneliness.
5. It is not possible to know the subjectivity of loneliness in mice when isolated. But in humans, loneliness is known to be a subjective phenomenon. For example, I could be alone at home and still not feel lonely, or I could be in a crowded place like a market and still feel lonely. So, inducing a subjective experience of isolation is a challenging task. To address this, socially well-connected participants were taken into the study, and they were kept away from any social interaction or stimulation like novels or even the internet. Questionnaires were used as well to assess the subjective experience of isolation and social craving.

Results:

The results obtained and their interpretation are:

1. Effect of isolation and fasting on social and food cravings: According to t-test values, participants reported social craving, increased loneliness and discomfort as well as decreased happiness compared to before social isolation. Similar was reported after fasting, i.e., food craving, hunger, discomfort, and decreased happiness were reported according to t-test values. Therefore, the absence of both forms resulted in craving, along with discomfort and decreased happiness. It was also noted that the standard deviation of craving ratings was more for social than for food. This implies more variability in the level of craving for social interaction among participants than for food craving. In my opinion, this represents the subjective nature of loneliness and disparities in the effect of isolation among people.

2. Comparison of SN/VTA activity between isolation and fasting days: In primates, craving for food and addiction, which are kinds of aversive motivation, have been found to be neurally represented in substantia nigra pars compacta and ventral tegmental area (SN/VTA). Since most of the SN neurons are dopaminergic (DA), fMRI BOLD signals from this area represent activity in the DA neurons. In this SN/VTA region, during the CIC paradigm task, it was found that activity in response to food cues was higher after fasting than after isolation. However, activity in responses to social cues after isolation was not significantly high compared to after fasting. Hence, SN/VT responds to motivationally noticeable deprived cues.

3. Comparison of SN/VTA activity between deprived days and baseline day: Activity in SN/VTA as a response to food and social cues after fasting and isolation was compared with responses to the same cues on the baseline day. It was expected that deprived cues would result in increased activity in response to the respective cue on the day it was deprived of. Instead, decreased activity in response to the non-deprived cue was found.

4. Activity in midbrain ROI: Activity in response to food cues is reported higher after fasting and significantly lower after isolation than baseline. This suggests a mechanism governing social craving in humans similar to that in mice, where DA neurons encode an aversive motivation state that motivates social interaction during loneliness.

5. Relation between self-reported cravings and activity in SN/VTA: Self-reported cravings were measured in two ways: during the CIC task in response to the craving cues (Craving_CIC) and as a questionnaire before the participants went for the scan (Craving_Q). It was found that for food cues after fasting, activity in SN/VTA was positively correlated with the Craving_CIC but not with Craving_Q. For social cues after isolation, activity in SN/VTA was positively correlated with Craving_Q, but not with Craving_CIC. So, even though the self-reported cravings were correlated with SN/VTA activity, the correlation was significant with respect to different conditions of self-reporting.

6. Similarity in activation patterns while feeling lonely and hungry: Multivoxel pattern analysis was performed to test the similarity in the activity patterns in response to social cues after isolation and food cues after fasting. A classifier was trained on SN/VTA activity to food cues and control cues (flowers) after fasting. Interestingly, the classifier decoded social cues from control cues after isolation with a mean accuracy of 0.542.

The classifier was even able to perform the opposite decoding as well, i.e., it decoded food cues from control cues after isolation with a mean accuracy of 0.545. Additionally, a representational similarity analysis was performed to alleviate the concern that the classifier’s generalization was solely based on learning of control cues and not the food or social ones. It was found that the pattern of activity in SN/VTA in response to social cues on isolation day was more similar to that in response to food cues on fasting day than on baseline day.

1. Correlations with chronic loneliness: The magnitude of SN/VTA activity in response to social cues after isolation varied among participants. This was found to be directly proportional to the self-reporting craving measure. It is reasonable to expect that this variability in responses reflects the pre-existing size of participants’ social networks or the presence of chronic loneliness. The authors found that this is, in fact, true. Participants with higher levels of chronic loneliness not only reported less craving for social cues during the CIC paradigm but even showed diminished activity in the SN/VTA region.

2. Activity in striatum: Striatum receives the majority of projections from midbrain dopaminergic neurons. The authors used a mixed-effects model to test each of the conditions against each type of cue for all sub-parts of the striatum, i.e., nucleus accumbens (NaCC), caudate nucleus (Ca), and putamen (Pu). Two results have been reported: a. Activity in response to food cues (compared with control cues) was higher after fasting (compared to after isolation) in NAcc, Pu but not in Ca. b. Activity in response to social cues (compared with control cues) was higher after isolation (compared to after fasting) in Ca but not in NAcc, Pu. This is an exciting result because, unlike in the SN/VTA region, where both social and food craving activity was similar, they were dissociable in the striatum.

3. Activity in whole-brain regions: Authors reported selective activity in response to food craving in the NAcc, ACC, periaqueductal gray, and amygdala. For social craving, activity was reported in Ca, OFC, and dorsomedial prefrontal cortex.

Critique:

• Participants with pre-existing chronic loneliness showed diminished SN/VTA activity in response to social cues (Point 7 in Results). This could be because of chronic isolation, probably leading them to social withdrawal. However, the causal connection underlying this correlation needs to be investigated further, as even low sensitivity of the SN/VTA regions could cause chronic loneliness, making it a consequence. • Craving was measured using passive viewing of cues and self-reporting, rather than a direct measure of the level of motivating behavior, it reflected. • The study was conducted with a small number of participants, and moreover, most of them were young adults. It has been seen that often, older people are more affected by loneliness. Combined with cultural differences across countries, this study cannot be generalized.

Future Directions:

• Since correlations between SN/VTA activity and self-reported cravings were significant w.r.t. to different conditions of self-reporting for social and food cravings (Point 5 in Results). Thus, changes in subsequent behavior as predicted by SN/VTA activity should be investigated. • The study should be conducted, taking participants from the older population to test the study results’ generalizability. • To understand the dynamics of how social craving and motivational activity are affected by the social environment, subsequent studies must be performed for each participant over more extended periods. • Lastly, it needs to be investigated which amount and kinds of social interaction (physical meeting, WhatsApp texts, etc.) can fulfill this social craving response.

Reference Tomova, Livia, Kimberly L. Wang, Todd Thompson, Gillian A. Matthews, Atsushi Takahashi, Kay M. Tye, and Rebecca Saxe. 2020. “Acute Social Isolation Evokes Midbrain Craving Responses Similar to Hunger.” Nature Neuroscience. https://doi.org/10.1038/s41593-020-00742-z.