Posted comment on ´Distinctive roles of medial prefrontal cortex subregions in strategic conformity to social hierarchy` written by D.Kim, J.Kim and H.Kim and published in Journal of Neuroscience 2023 vol 43(36) p.6330 doi 10.1523/JNEUROSCI.0549-23.2023
SUMMARY
Kim, Kim and Kim looked at the different roles of the dorsal and rostral medial prefrontal cortex (mPFC) during a type of social engagement where an individual`s behaviours and decisions were aligned to the expectations of others who had been deemed of higher (termed ´superior partners`) or lower social (termed ´inferior partners`) standing.
Kim, Kim and Kim`s study involved 45 participants (24 females, mean age 22.5) performing a preference rating task where the preferences for novel stimuli were influenced by the preferences of others. The participants were first informed that they were to play an online game with two other unknown participants who were not in the same room. The participants were told they would take part in three computer-based phases of the task which were: social hierarchy manipulation, preference rating task and a modified dictator game. Before beginning the social hierarchy manipulation and preference rating task, the participants saw on a screen the two other participant/partners appearing to access the online tasks. This was carried out to convince the participants that their experimental ´partners` were real.
The first part of the experimental procedure was to manipulate the participant´s social hierarchy. This was done by using two simple perceptual tasks, designed so the participants were not able to assess their real performance. Although the participants were informed that their performance would dictate their hierarchy in the subsequent experiments, all participants were assigned an intermediate rank between their imaginary partners. They were also informed before the task that their final decisions in the public condition trials would be viewed by all partners immediately on the completion of the preference rating task, but their final decisions in the private condition trials would remain anonymous.
The experimental procedure required participants first performing 10 practice trials before entering the fMRI scanner. When in the scanner, they were asked to rate 120 fractal images presented on a log-on screen for the main preference rating task. The preference rating task trials consisted of five distinct phases: first – presentation of a black fixation cross followed by a fractal image presented on a 4-point Likert scale where the participant had to evaluate and rate the likeability/´dislikeability` of the image (termed the ´first preference rating`); second, self-feedback stage – visual presentation of the participants ratings (rating 1 or 2 – thumbs down; rating 3 or 4 – thumbs up); third – display of the either the superior`s or inferior`s preference rating of the image with those ratings being determined to be evenly distributed relative to the participants decisions (termed ´other-feedback`); fourth – the participants were shown a symbol to remind them they were either in the public condition (symbol – eye in magnifying glass) or the private condition (symbol – image of padlock) (termed the ´observation cue`) ; fifth – repeat presentation of the fractal image to promote re-evaluation and rating by the participant (termed ´second preference rating`). In total there were 15 trials organised into the following 8 conditions (key – Hi – superior/higher; Lo – inferior/lower; Ic – incongruent condition; Co – congruent condition; Pu – public condition; and Pr – private condition): HiIcPu, HiCoPu, HiIcPr; HiCoPr, LoIcPu, LoCoPu, LoIcPr, LoCoPr. After the tasks were completed, the participants filled in questionnaires measuring individual differences, were asked a few verbal interview questions and then debriefed about the deceptions regarding the absence of real human participants and modified dictator game. This modified dictator game, which did not take place, had been included in the initial instructions to emphasize and enhance the external validity of the hierarchy manipulation.
The behavioural data was analysed according to pre-determined rules. For better representation of the preference rating change as the result of conforming behaviour, preference ratings were binary coded (ie. 4 – strongly like plus 3 – like were coded as ´1` and 2 – dislike and 1 – strongly dislike coded as ´0`). The preference change in each trial was calculated and ranged from 0 to 1. Then the mean preference change score for each condition was assessed by repeated measures ANOVA (2 times – social hierarchy, superior and inferior by 2 times partner`s reference, incongruent and congruent by 2 times observation, public and private). Statistical significance was assessed by paired sample t tests. Kim, Kim and Kim also calculated the greater likelihood of conforming to a superior`s partners opinion under both public and private conditions (ie. individual tendency of strategic conformity-SC – calculated by performing ((HiIcPu – HiCoPu) – (LoIcPu – LoCoPu)) – ((HiIcPr – HiCoPr) – (LoIcPr-LoCoPr)).
FMRI was carried out during the behavioural task and the images pre-processed by correcting for slice timing and head motion, normalised and spatially smoothed. The data was then processed according to a first level general linear model analysis (GLM) to estimate neural responses to the experimental conditions and a second level group analysis to test whether a strategy was established by the participants. The GLM included 16 regressors of interest which were: first rating of all images with the participant`s response time as a boxcar function; four regressors representing the other feedback onset of the partner`s hierarchy and preference for each condition; 8 regressors representing the onset of the second rating presentation with a parametric modulator reflecting choice behaviour (1 – switch or stay, minus 1 for other); the onset of the decision; and 6 motion parameters included as regressors of no interest (eg. one regressor represented the decision phase with the button press in order to reduce the noise of the pressing movement). Beta estimates were extracted and analysed by 2 by 2 ANOVA.
The second level group analysis instead involved subject-specific contrast images entered using a one sample t test. The correlation used was the contrast of the social hierarchy times partner preference interaction and testing of the group effect in order to find the region that corresponded to the partner`s hierarchy and preference information. Kim, Kim and Kim also defined the contrast of the three-way interaction according to the formula ((HiIcPu – HiIcPu) – (LoIcPu – LoCoPu)) – ((HiIcPr – HiCoPr) – (LoIcPr – LoCoPr)) and in addition aimed to identify the brain region that responded to the interaction of partner preference times observation in the superior condition and interaction of social hierarchy times observation in the incongruent condition at the time of observation cue display. Voxelwise multiple regression analysis was also used.
Kim, Kim and Kim then went on to report their results, which they divided into two groups: behavioural results and neuroimaging results. The first set of behavioural results related to whether or not individual preferences would be changed to align with those of other people particularly in public situations. The ANOVA results based on binary coding showed that the changes in partners` preferences to the presented fractal images were influenced by the social hierarchy of the partner, observation and the partner`s preference. There were significant two-way interaction effects (social hierarchy times observation; observation times partner`s preference) and a three-way interaction effect (strategic conformity). When the results were analysed using participants` choices as continuous variables (ie. absolute values of differences between second and first ratings) then a significant three-way interaction effect was observed (2 – social hierarchy, superior and inferior times 2 – observation, public and private times 2 – partner`s preference, incongruent and congruent). There were also, significant two-way interaction effects (social hierarchy times observation; social hierarchy times partner`s preference; observation times partner`s preference). It was also found that the changes in participants` preferences were influenced by the social hierarchy of the partner, the observation condition and the partner`s preference. Therefore, in general the results of both binary coding and continuous variables were similar.
Using the binary coding results and in order to assess behavioural responses between pairs of conditions, Kim, Kim and Kim then conducted post-hoc paired sample t tests. Participants were found to be more likely to change in their preferences in line with those of the superior partner in public conditions in the incongruent condition. However, in the congruent condition Kim, Kim and Kim found that there was no significant difference between the inferiors` and superiors` decisions. They also found that with the incongruent conditions, participants were more likely to change their preference following the superiors` or inferiors` opinions.
Using the binary coding results for the private conditions, Kim, Kim and Kim found that participants were more inclined to conform to the superior partner`s opinion than the inferior in the incongruent condition. The participants were therefore, more likely to change their preference following the superior`s and more likely to change their preference following the inferior. However, in the congruent condition, the authors found no significant difference in preference rating changes between superior and inferior trials and the participants were less likely to change their preference following the superior or inferior partner.
Using post hoc paired sample t tests, Kim, Kim and Kim found that participants were more likely to consider information on social hierarchy and observations in the incongruent condition. The participants were more likely to change their preference in the superior under the public condition compared to the private. They were more likely to change their decision to follow the superior rather than the inferior under the public condition. Under the private condition, then participants were more likely to conform to the superior than the inferior. In the case of the congruent conditions then the participants were more likely to change their preference when the superior`s rating was presented under the private condition than under the public. Kim, Kim and Kim found when they investigated the difference between baseline and preference changes in the congruent conditions that participants were more likely to change their preferences under the private conditions when the superior`s rating or the inferior`s rating were presented. In the case of public conditions, the participants were more likely to change their preferences when the inferior`s rating was presented whereas there was no difference when the superior`s rating was presented.
The next set of behavioural results given by Kim, Kim and Kim related to reaction time. The authors found that reaction times were longer in the incongruent conditions than in the congruent conditions and also that the reaction times were significantly longer within the incongruent trials when the participants were presented with the superiors´ preferences compared to the inferiors.
The next set of behavioural results reported related to the rating discrepancy and the author`s investigation into whether the participants conforming behaviours were value-based (cost-benefit) or rule-based decisions. When participants initially chose a rating of either 2 (dislike) or 3 (like) for the incongruent trials the authors assigned a value of 1. Ratings of 1 (strongly dislike) or 4 (strongly like) were assigned the value of 2. The authors found that for these trials, there was a significant negative correlation between rating discrepancy and reaction time. This was not observed for the congruent trials. Also, a significant negative correlation was observed between rating discrepancy and reaction time in the superior conditions, but no significant correlation was seen in the inferior conditions. It was also observed that participants made choices with a large rating discrepancy (either 1 or 4) in 25% of the trials in superior conditions (336 from 1350 choices) with 12% (39 of 336 choices) exhibiting conforming behaviour. In 75% of the trials, the participants made small rating discrepancy (2 or 3) in the superior conditions (1014 of 1350 choices) with 32% (327 from 1014) of choices exhibiting conforming behaviour. Kim, Kim and Kim concluded that participants were more likely to change their preferences when the rating discrepancy was small and the superior`s rating was presented. Based on the results of correlation analysis, Kim, Kim and Kim said that participants were capable of quick decision-making to disagree with the superior`s opinion when the rating discrepancy was large but when the difference was small then an additional cognitive process would be required. This suggested that the participants experienced heightened conflict indicating that conforming behaviours were not solely based on simple rule-based decisions but on complex value-based decision-making which would take into account various factors such as the disparity between initial choice and the choice made by others.
The final set of behavioural results related to correlation analyses to assess the relationship between social dominance orientation (SDO) and fear of negative evaluation (FNE) and social conformity (SC). Previous work had shown that the extent of the latter could be predicted by the former SDO and FNE. Kim, Kim and Kim found that there was a significant positive correlation between FNE and SC scores, but not between SDO and SC scores.
Kim, Kim and Kim then went on to describe their neuroimaging results. The first set of results related to neural region activity in response to congruency with a superior partner`s opinion. The authors found that the main effect of social hierarchy at the other-feedback onset and therefore, examination of the contrast maps of social hierarchy (superior vs inferior) times partner`s preference (congruent, incongruent) interaction at the other-feedback display was said to provide information about the neural regions that would respond differentially to different opinions. Kim, Kim and Kim reported significant activity in the vmPFC and the VS. Further examination using post hoc paired sample t tests showed that the activation was significantly greater in both areas (ie. vmPFC, VS) when participant`s preferences were congruent with the superior partner in comparison to the inferior partners. It was also reported that both the vmPFC and VS exhibited greater responses when the participants opinions were incongruent with those of the inferior partner compared with the superior. This also occurred when the preferences were congruent versus incongruent with those of the superior partner. These results were said to indicate that the participants experience heightened reward when the preferences were congruent with those of the superior partners but experience reduced reward when their preferences were incongruent.
The next set of neuroimaging results related to the identification of the neural regions linked to increased conformity to the social hierarchy in the public condition compared to the private. Contrast maps of the three-way interaction at the time of observation cue display were examined by the authors. They found no significant clusters and as a result looked at the contrast of observation times partner`s preference interaction at the superior condition. In this case, significant clusters in the dmPFC and the vmPFC were found. Further post hoc analysis showed that the dmPFC activation was significantly greater when the participants` preferences in the private condition were incongruent versus congruent and when participants` preferences were incongruent in the private versus public condition. In the case of the vmPFC, like dmPFC activation was greater when the participants` preferences were incongruent in the private versus public condition. However, in the case of congruent condition then the vmPFC showed a greater increase in activity when participants` preferences were congruent versus incongruent in the public condition. It was also significantly greater when participants preferences were congruent in the public condition than in the private condition and when their preferences were incongruent in the private condition than in the public condition. Therefore, Kim, Kim and Kim concluded that both the vmPFC and the dmPFC are involved in the additional cognitive processes associated with strategic conformity particularly when participants` preferences were incongruent with those of the superior partners in the private condition. An investigation of whether the neural signals in the dmPFC and the vmPFC in the superior condition were involved in differentiating social information in the inferior condition (interaction of observation times partners` preferences) saw that there were no significant interaction effect in either region. The dmPFC only exhibited a greater response when the opinions of the inferior partner were congruent in the public versus the private condition.
The next set of neuroimaging results related to Kim, Kim and Kim`s investigation into identifying the neural signatures of individual differences in strategic conformity to the social hierarchy in the public versus private condition at the time of other-feedback display. The authors found when they ran a voxelwise multiple regression analysis that there was a large cluster in the rmPFC which indicated a significant positive correlation between the SC and the control maps of social hierarchy times partner`s preference interaction at the time of the other-feedback display. By using a median split in order to divide the participants into high and low SC groups, it was found that the high SC group exhibited significantly greater rmPFC activation when the participants` preferences were congruent versus incongruent with those of the superior partner. This observation was found to be reversed for the inferior partner which showed a significant two-way interaction effect. Further analysis using post hoc t tests indicated that the rmPFC activation in the high SC was significantly greater when participants´ preferences were congruent with those of the superior compared with the inferior partners. This also occurred when partipants´ preferences were congruent versus incongruent with those of the superior partner. When participants` opinions were incongruent with those of the inferior compared to the superior partners then the rmPFC activity was also observed to be significantly greater. The low SC group however, showed no significant two-way interaction effect. It also exhibited no significant difference in any pairs of conditions. Therefore, Kim, Kim and Kim concluded that the rmPFC activity reflected the additional cognitive processes involved when considering others´ opinions during decision-making.
The next set of results related to the identification of the neural signatures of individual differences in SC to the social hierarchy in the public versus private conditions when the observation cue was displayed. In this case, activities of the dmPFC and the TPJ were positively correlated with SC scores in the contrast of social hierarchy times observation interaction. Further post hoc analyses showed that the high SC group showed significantly greater activities in both dmPFC and TPJ regions in the private versus public condition when paired with the superior versus inferior partners. However, the low SC group showed no significant interaction between social hierarchy times observation which indicated no significant differences between any pair of conditions. Kim, Kim and Kim therefore, concluded that the increased activity in the dmPFC and TPJ were implicated in the additional cognitive processes required for considering others´ hierarchical positions and the observational context which facilitates the detection and resolution of conflicts in the private condition.
Having reported the results of their behavioural and neuroimaging experiments, Kim, Kim and Kim then proceeded to discuss their findings. They began by summarising their results. From a behavioural perspective they found: participants were more likely to conform to others in higher versus lower social positions especially when participants believed that their decisions were visible to those in the higher social positions; and participants demonstrated a propensity to retain their opinions in congruent conditions especially when their opinions were visible to the superior partner potentially mirroring the motivation underlying the conforming behaviour observed in the incongruent conditions. From a neural perspective, the authors found: the vmPFC and the VS showed increased activities when participants` preferences were congruent versus incongruent with those of the superior compared with the inferior independent of the behavioural results; the rmPFC showed increased activity when the initial decision was congruent with that of the superior partner and incongruent with that of the inferior partner, hence indicating a goal-dependent valuation; and the dmPFC with the TPJ showed increased activity in the private condition versus the public condition when paired with the superior partner versus the inferior. Therefore, activity patterns in the rmPFC and dmPFC were observed only among those with a higher tendency to conform to the social hierarchy under social observation unlike the vmPFC. Based on this Kim, Kim and Kim concluded that their findings supported the mPFC model for social observation where the ventral to dorsal gradient of the mPFC subregions is hierarchically structured. This means it can exert allostatic control over internalised social responses by adding more information from the external environment.
Although evidence of activity in mPFC in social decision-making where decisions may be made with consideration of other peoples` opinions, preferences, attitudes of the majority or social status had already been known, Kim, Kim and Kim then went on to discuss how different mPFC subregions could play functionally dissociable roles in strategic social conformity in particular. The view that social conformity to others is a type of reward-seeking behaviour is supported by the activities observed in the vmPFC and the VS, known to be two key neural hubs of the reward processing network. Kim, Kim and Kim said that their findings provide additional support, ie. activities of the vmPFC and VS increased when the participant`s initial choice was congruent versus incongruent with those of superiors and decreased when the participant`s initial choice was incongruent with those of superior versus inferior partners across all participants irrespective of whether they conformed to the superior partners` opinions. The authors went on to say that people have generally a highly internalised reward value associated with social hierarchy and therefore, are likely to have a feeling of gratification when their opinions are congruent with others who they regard superior, but not with those regarded inferior. Therefore, this internalised reward value of social hierarchy could serve as a major driving force for subsequent conformity to a superior partner. Bearing this in mind, Kim, Kim and Kim suggested that since vmPFC activity is associated with visual attention then eye-tracking techniques could be used to investigate the vmPFC response to social strategic conformity.
Kim, Kim and Kim then went on to describe the role of the rmPFC located immediately dorsal to the vmPFC. This area is known to track and update information about the social hierarchy and promotes self-enhancement behaviour in a context-dependent manner. These known roles appear to be supported by Kim, Kim and Kim`s studies since the rmPFC showed increased activity when an initial decision was congruent with a superior partner and incongruent with an inferior partner. This occurred only with individuals who demonstrated a high SC. Therefore, Kim, Kim and Kim suggested that their findings contributed to the knowledge about social decision-making by proposing that rmPFC activity reflects additional consideration of social hierarchy and congruency of opinions to compute the value of the decision to conform to superiors as a means to promote one`s relative social position.
Kim, Kim and Kim then went on to discuss the role of the dmPFC which they found, along with the TPJ, showed increased activity in the private versus public condition when paired with superior versus inferior partners. The rmPFC was found to compute the value of the decision to conform to a superior partner before the observation cue appeared and hence, conforming decisions become default decisions when the public condition cue is displayed whereas extra control is required to override the default decision when the private condition cue is displayed. Therefore, people with higher SC scores in the private condition would face stronger conflict between the desire to favourably impress others of a higher-ranking position and to maintain their initial opinion for consistency. The increased activity in the dmPFC and TPJ in the private condition may be involved in detecting and resolving the conflict by more careful consideration of external information, eg. evaluation of more details of visual features of the images, mentalising process.
Kim, Kim and Kim concluded their discussion by looking at social conformity with respect to minimising metabolic costs. This came about from previous hypotheses that human social behaviours are best understood as products of the neural process of seeking optimal strategies to minimise prediction errors arising from the mismatch between the internal model of the world and physiological and behavioural consequences and to regulate one`s interaction with the social environment at efficient metabolic cost. Social conformity was described as one example of this since efficient metabolic expenditure would come from making an individual`s social environment more predictable by learning what others would expect in a given social context. Increased conformity to others in a higher social hierarchy by social observation would then be regarded as an efficient neural strategy for minimising metabolic costs and uncertainty. However, Kim, Kim and Kim found no additional neural activity at the time of the observation cue which suggested a public versus private condition when the subject was paired with a superior partner. Kim, Kim and Kim suggested that therefore, conformity to the social hierarchy may be engaged without additional metabolic costs under social observation probably because the value of the conformity to the social hierarchy had already been computed by the vmPFC, VS and rmPFC even before the cues of the public or private conditions. However, Kim, Kim and Kim said that in their studies they did find additional activity in the dmPFC and the TPJ which may be responsible for overriding the value of the default decision (ie. conformity) because the social pressure to conform to a superior partner had been lifted. In this case, then the motive to maintain consistency of opinion which could be an alternative way of promoting self-esteem and social position would win the competition and initiate a value computation for a choice against conformity at the expense of extra metabolic costs.
Kim, Kim and Kim summarised their study findings by confirming that they had found distinct subregions of the mPFC differentially involved in conformity to social hierarchy under social observation which supports the mPFC model for social valuation. They suggested that the ventral to dorsal gradient of the mPFC means that it exerts increasing allostatic control over internalised social responses by adding more information from the external environment. This would be a means of minimising the metabolic costs required. They concluded that further investigation would be needed to elucidate whether this type of social conformity was common to all cultural contexts.
COMMENT
What makes this article interesting is that it continues the investigation into the specific functioning of different medial prefrontal cortex (mPFC) regions involved in decision-making. In Kim, Kim and Kim`s investigation they looked at activities of specific areas of the mPFC (dorsal, ventral rostral) and the ventral striatum (VS) corresponding to their study participants changing decisions in order to agree with the opinions of others. This is not the only one reason why decisions may be changed, eg. changes in environmental conditions, changes in goals, but it is important in everyday social behaviour, debates and discussions. Another subtlety of the experimental set-up included was that the participant`s decision or change in decision would either be made known to others or kept anonymous. Therefore, this addition put the participant under pressure of agreement or disagreement to social group opinion.
Kim, Kim and Kim`s study involved a simple visual task where there were two decision-making stages. The first decision involved the participants deciding the level of ´likeability` of fractal images of people (First Preference Rating) followed by a reaffirmation of that decision (thumbs up or down – Self-feedback stage). The next stage involved the participants seeing the opinion of another person (the partner) to the same image (the Other Feedback). Here, Kim and team altered the partner to be either a person considered by the participant to be of higher social authority/status (the superior partner) or lower social authority/status (the inferior partner). Once the participant had seen the partner`s opinion, they were then told whether they would be made visible (public condition) or whether they would remain anonymous to the partner (private condition). The participants were then given the opportunity either to change their mind about the rating they had previously given or to maintain their original decision (Second Preference Rating).
From a behavioural perspective, Kim, Kim and Kim found that their participants were more likely to change their own opinions about the images to match those of their partner when they thought that the partner was of higher authority/social status. This occurred whether their decision was ´public` or ´private`. If the partner was considered to be of lower authority/social status and the participant disagreed with their view then opinion was not changed. Therefore, Kim, Kim and Kim said that participants had demonstrated social conformity to those of higher authority/social status. The authors then investigated firing activities in the mPFC and VS regions associated with this decisional behaviour. Using fMRI, Kim and team found that neural activity was found in distinct subregions of both areas according to behaviour. For example, there was increased activity in the mPFC when the participant`s opinion aligned with that of the superior partner (ie. increased conformity to social hierarchy). This split into increased activity in the ventral mPFC when the participant`s preferences aligned with those of superior partners plus there was also increased activity in the rostral mPFC (rmPFC). There was also increased activity in the rmPFC when the participant did not conform to the decision of the inferior partner. The authors interpreted this as an indication of goal-dependent valuation in decision-making. Activity in the dorsal mPFC (dmPFC) was also shown to be increased, but in this case the increase was only observed in private conditions with a superior partner. This activity was also only observed with participants who had a higher tendency to conform. Therefore, Kim, Kim and Kim concluded that their findings showed that in their case of social decision-making then there was a hierarchical allostatic regulation model of mPFC function for social valuation. They also suggested that strategic conformity was a way to minimise metabolic costs.
The comment to this article posted here concentrates on several of the PFC neural areas required for successful performance of the decision-making involved in Kim, Kim and Kim`s task. Social decision-making has been found to engage the same areas as non-social decision-making, for example in relation to the PFC (the area focused on by Kim and team) then vmPFC and dorsolateral PFC (dlPFC) activities remain the same but TPJ and pre-SMA activities change (Kolling). Therefore, the ´rules` applied to the decision-making process and the neural activities observed are assumed to be the same independent of whether the task is non-social or social. With regards to the task given to the participants by Kim and team, decisions have to be made by their experiment participants at two stages and these decisions are termed the First and Second Preference Ratings. When the decision-making processes are considered from a system perspective, then the first occurrence (the First Preference Rating) would require probably System 1 decision-making (Evans) having the characteristics of being rapid, automatic and probably capable of being performed in parallel to other tasks. This can be assumed because this part of Kim´s task probably relies on the emotional system and personal values only since each visual image is assessed according to personal pleasure/likeability standards. The Second Preference Rating would probably require System 2 decision-making described as having the characteristics of being slower, sequential and working memory-dependent, ie. requiring information processing (Evans). The major difference between the two systems is the involvement of different PFC regions which associates the general decision-making process to Kim and team`s article, ie. System 1 – ventral medial PFC activity and System 2 – right inferior PFC activity.
From a cognitive demands perspective, this comment follows the general stages of decision-making proposed by Buzan. This PISCO system consists of several stages: Purpose – definition of goal; Input – report of all relevant information; Solutions – proposal of a number of solutions to achieve the goal and dependent on known or given strategies; Choice – selection of appropriate or most appropriate option by means defined by the individual; and Operation – resulting action (behavioural or cognitive). Feedback on the whole process (eg. success, performance level, measure of conflict) follows. With regards to Kim`s task, the first stage, Purpose is given by the experimenters and can be defined as ´report likeability of image` for the decision-making task linked to the First Preference Rating, but for the Second Preference Rating, then the Purpose would be defined as ´if my opinion does not match that of my superior partner, would I be better ´served` by changing it?`. Two things should be noted here. The first relates to the word ´served`. In this context, ´served` would mean the personal status, beliefs, feelings, values of the participant who would probably seek to maximise their status, seek parity with the partner, reduce self-deprecation etc. Therefore, it is assumed that an attempt is made by the participant to maximise or retain his feelings of self-worth. The option of indifference is not catered for. Bearing this in mind, the goal for the Second Preference Rating then changes because not only is the partner either of superior or inferior regarded status there is also the additional factor that the partner may not be able to attribute the decision to a particular person (private). This would make the option of changing irrelevant because the incongruent decision would not be associated to the specific participant and therefore, there would be no effect on status because of any decision made. If the partner could attribute the decision to a particular person (public condition), then the participant has to consider the social impact of not changing an incongruent decision. Therefore, the Second Preference Rating relies on a decision to basically ´switch or stay`.
The differences in Purpose/Goal definition for the First Preference Rating and Second then translates into different second stages, that of Input. For the First Preference Rating, the input relates to personal likes/dislikes recorded in terms of values via activity of the emotional system. This Input relies on recall of these values and is likely to be automatic. However, the Input for the Second Preference Rating is structured according to the internal question of ´what would happen if I did not follow a superior partners value`. This would then demand access to previous experience and stored knowledge about social rules plus consequences of failing to comply to the wishes of others (ie. includes personal feelings plus actions of others and Self).
According to Buzan`s PISCO stages of decision-making the next stage ´Solutions` is also different for the First Preference Rating and the Second. In the case of the First Preference Rating then the options available in the ´Solutions` stage are simple and involve accepting the values given by the emotional system for each facial image presented. The ´Solutions` of the Second Preference Rating instead involve the construction of appropriate strategies appropriate for this particular task. There are many different strategies and individuals tend to favour some more than others dependent on situation. For Kim`s task the likely strategies are ´look at aims, goals and objectives`, plus because of the nature of the task ´other people`s views` (this would mean the partners) and ´consequence and sequel` (this would mean ´what would happen if I did not switch to agree` and ´does my decision being made public to the partner make a difference to me`). The absence of no direct experience of this type of scenario or use of strategy would mean that additional processing and imagination would be required. This is possible and imagination is described as the experience of novel situations during which new associations are constructed by invoking multiple independent memories in the hippocampus and mPFC (Barron). Again, this reinforces why mPFC activity was chosen by Kim as an area to study. It is also confirmed by the view that the formation of strategies requires prospection since model-based choices are found to be associated with neural activities linked to this type of cognition and memory at this time (Doll).
Buzan`s Choice stage is thought to rely on comparison of strength of firing of competing groups. Therefore, for the First Preference Rating, the choice is relatively simple as the reactivated opinion based on the given emotional system value. This Choice stage for the Second Preference Rating requires the choice to suit the purpose/goal. Therefore, the option is chosen by comparing the strengths of firing of different groups representing an acceptable solution from the emotional angle (value and priority) according to agreement of opinion conflict by conforming. Therefore, for some individuals the decision would involve the emotional ´cost` of conforming plus a fear of status decline if a change in opinion is refused (risk) and increased kudos if the decision agrees. It is clear that the success at this ´Choice` stage depends on individual values, individual regard to personal status in the eyes of others (group conformity), motivation as well as efficient and accurate Input and Solutions stages.
The final stage of Buzan`s decision-making process (the ´Operation`) stage requires an action and in both the First and Second Preference Ratings then it involves the physical action of pressing a button to indicate choice (First) and the physical action of button pressing to change the original decision (Second). There may also be a Feedback stage where the outcome is acknowledged, monitored and future decisions adapted accordingly (Wokke). This probably does not occur in Kim`s experiments since the images presented are independent of each other and therefore, learning across trials is not possible.
The other part of Kim, Kim and Kim`s experiments involved looking at neural activities in particular areas of the mPFC in the responses to their decision-making task. As given above in the description of the processes required to carry out successful decision-making according to Buzan, it can be seen that a number of mechanisms are required in decision-making (eg. attention, working memory, memory recall) and hence, a number of different brain areas. For example, decision-making is linked with activities in parietal cortex, basal ganglia, motor structures (Kiani), PFC (decision monitoring – Tsiyomoto; expected reward from DA neuron activity in PFC – Floresco, Redgrave, Niv) and cingulate cortex (unexpected reward CC – Niv).
Since Kim, Kim and Kim`s experiments concentrated on the PFC, this comment restricts to this particular region in particular. The PFC plays an important role in decision-making (general functions associated with PFC include working memory, emotions, attention, central executive) and activity is dependent on different areas (Brodmann areas), activation and responses (Wang, Raisin) and shows location and functional dependency. For example, in the case of different activation PFC relies on the mechanisms relating to many neurotransmitters such as dopamine (DA) and serotonin (5-HT). In the case of dopamine, PFC DA regulates various executive functions such as attention, working memory with DA1R and DA2R activities being reported (Floresco). It has been found that DA population activity and not reward during reference trials determines choice (Talbot) and DA neurons signal the occurrence of unexpected rewards (Redgrave). It has also been found that DA2 receptors in the PFC enable flexible decision-making and DA1 receptors promote persistence in choice biases (Jenni). Disruption of the PFC to basal amygdala pathway (PFC-BLA) leads to impaired adjustments in decision biases in response to changes in reward probabilities and disruption of DA1 modulation of the PFC to the nucleus accumbens reduces risky choice, increases reward sensitivity and increases sensitivity to reward omissions (Jenni). In the case of 5HT, then depletion of tryptophan (precursor for 5HT) leads to a lack of discrimination of rewards independent of risk (Rogers).
This variability in neurotransmitter functioning and hence, pathway activity results in different areas of the PFC being involved in different functions. With regards to decision-making, for example: the left PFC is associated with the left temporal cortex and ACC in the function of planning with practice reducing activity (Anderson) and the right inferior associated with the slow, sequential working memory-dependent System 2 decision-making (Evans). The lateral PFC is reported to be involved in strategic control (MacDonald) and in particular the dorsolateral PFC (dlPFC) divided with the right linked to plan generation and the left, plan execution. It has also been found that a larger proportion of neurons are activated in low motivational conditions in the dlPFC than in the ACC and onset of this activity is significantly earlier in the dlPFC than the ACC (Amemori). DlPFC activity also appears to be involved in decision-making because it encodes both task relevant and task irrelevant features but only task relevant information is found to be encoded congruently with choice signals (Donahue).
However, Kim, Kim and Kim looked at the medial PFC (mPFC) areas in particular and this area was chosen because the nature of their task required decisions to be based on personal values. The medial PFC is known to play a significant role in this as well as attention which is also a more general decision-making requirement. For example, in the consideration of Kim`s task according to the processes put down by Buzan, as given above it can be seen that Kim`s task requires value-based decision-making. In general, the First Preference rating would be a reflection of the participant`s own personal value and the Second Preference rating a comparison of this value to others. Therefore, variations in value can cause differences in decisions and future decisions if learning occurs. For example: inherent values may interfere with or facilitate the acquisition of new expected values during associative learning (Chien); updating of reward values may be required; and participants when feedback of failure is received then are found to rely more strongly on reward magnitude when choosing whether to exert greater effort to obtain larger rewards (Anand).
The prime function of the PFC and particularly mPFC is however, the calculation of value of reward (Wallis, Kringelbach) and therefore, with reference to Kim´s task, the value of agreement to the superior partner`s opinion. One example of evidence for the association of mPFC activity and value directly is that the activity in the dorsal medial PFC is modulated to upcoming action value and activity in this area determines behaviour and not reward (Morris). Therefore, it was concluded that decisions are generated elsewhere and conveyed to DA neurons which then play a role in dorsal medial area by affecting basal ganglia efficiency (Morris). Other research shows that choice when options have different reward values relies on communication from the dorsomedial frontal cortex to the subthalamic nucleus (STN) which allows more time to choose the statistically more rewarding option (Frank). Also, during individual choices, the OFC alternates between states associated with the value of two available options (neural assemblies) with dynamics that predict whether a decision is taken quickly or oscillates between the two (Rich). This would relate to Buzan`s Choice stage. Medial PFC activity is also observed when the Choice stage does not relate to the individual (or Self) but to others. For example, in the case of communication between the OFC and anterior cingulate gyrus (ACC) it is thought that the neurons in the monkey OFC predominantly encode rewards that are delivered to oneself and neurons in the ACC encode reward allocations to the other monkey, to oneself or both (Chang). This is supported by another study which relates directly to Kim, Kim and Kim`s task. For example, a difference between dorsomedial PFC and vmPFC in consideration of self-values and values of others has been observed (Zhao). In this case, research found that the neuropeptide oxytocin (OXT effect which can motivate individual to value the attributes of others in a more self-like manner – normal regarded as higher value – the ´endowment effect`) reduces mPFC responses (Zhao). OXT was found to increase the self-value effect and value of possessions of others by reducing activity in the vmPFC in the self-owned possession and increasing it in the mother-condition. A functional connectivity between dmPFC and the ventral striatum (VS) was also found to be decreased and between vmPFC and the precuneus (Zhao) although activity in the dorsal mPFC however, was seen to increase for others` possessions (Zhao). This shows an association between value, decision-making, Kim`s task and mPFC activity and particularly vmPFC activity, the focus of Kim`s experiments.
The vmPFC is known to be associated with System 1 rapid parallel, automatic, belief-based decision-making (Evans) and there is much supporting evidence of its involvement. For example: activity representing expected value is found in the vmPFC (Chien); increased vmPFC level of activity before a decision is made in certain circumstances and increasing regret shows activity in the medial OFC, ACC and hippocampus (Coricelli): vmPFC activity reflects both value comparison and confidence in the value comparison process (De Martino) and when the participants know they have to choose between two options then there is a strengthened pathway between the hippocampus and striatum (Murty); the OFC links events to reward values and represents contexts that guide memory retrieval (Favonick); and the OFC computes the subjective value of outcome (Wallis, Hornack, Lak). The physiological nature of value appears to be through GABA and glutamate concentrations in the vmPFC (Jocham) with value representations found to be organized through an anterior-posterior gradient (secondary-primary rewards) (Li, McNamee).
This activity in the vmPFC linked to value and reward also appears to be associated with vmPFC and ventral striatum (VS) connectivity and this was demonstrated by Kim, Kim and Kim. Kim saw in their experiments that congruent choices with both the superior and inferior partners led to activity greater in the vmPFVC and VS regions. These observations regarding increased activity in the VS are supported by the known roles of the striatum and sub-regions. The striatum consists of several areas, ie. dorsally – the caudate and putamen and ventrally – the nucleus accumbens with most areas associated with the emotional system and hence, values. The importance and functionality of the region are supported by the varied cell types, active neurotransmitters and connectivity to other brain areas. For example, nearly all cells (95%) of the region are medium spiny neurons and many neurotransmitters in these cells are active, eg. GABAergic and inhibitory or DA1/DA2. There are also large aspiny interneurons that are cholinergic (excitatory), but there are also many types of GABAergic interneurons (including tonically activated interneurons). From a connectivity perspective, corticostriatal pathways are ideal for information integration. For example, the striatum has two bilateral zones, ie. one in the caudate, the other the putamen with projections to the OFC, dlPFC and parietal areas (Jarbo). The cortical inputs and striatal outputs are organized to form three integrated cortico-amygdala-striatal systems (Cho). Other pathways also include input to the ventral striatum from the thalamus, hippocampus, entorhinal cortex and inferior temporal gyrus plus output to the rest of the basal ganglia with primary outputs of the VS for example to the ventral pallidum, medial dorsal nucleus of thalamus, globus pallidus and substantia nigra.
The structural versatility of the striatum region leads to its wide-ranging functionality such as motor and action planning, reward motivation and reinforcement and relevant to Kim`s study, value-based decision-making. Evidence from this comes from a number of sources, for example: the strength of white matter tracts from the vmPFC to the medial striatum which is top-down control is involved in the two modes of decision-making (Piray); and the striatum is known to play role in reinforcement learning specifically in encoding of teaching signal such as reward prediction errors with impaired coding of RPE and heightened DA turnover in striatum (Boehme).
More specifically, Kim, Kim and Kim looked at the functional connectivity between the vmPFC and the ventral striatum (VS). This would be observed because of the nature of the Kim`s decision-making task which is reliant on value assessment and updating. Support for this type of activity comes from a number of sources. For example, functional coupling of striatal areas to vmPFC during choice is reported and therefore, it was suggested that these regions are likely to act as a value comparator (Wunderlich) and large interindividual differences in value-guided control can be predicted by the structural integrity of white matter tracts from the vmPFC to the medial striatum (Piray). Variations in activity and connectivity between the areas may have effects on decision-making. For example, inherent values may interfere with or facilitate the acquisition of new expected values during associative learning (Chien). One study (Chien) having attractiveness of a facial picture as a proxy for inherent value of the cue and the picture paired with monetary win (congruent) or loss (incongruent), found that participants had faster response times and faster learning rates for value-congruent outcome pairings (Chien). This study saw that activity in the VS was linked to both learning rate and reward prediction error and activity representing the expected value was in the vmPFC (Chien). This study would correlate with Kim`s social conformity task. Also, incidental negative emotions unrelated to current tasks such as background anxiety was observed to strongly influence decisions (Engelmann). In this case, activities in the vmPFC and VS showed a marked reduction in neural coding to expected subjective value (ESV) of risky options (Engelmann). A marked reduction in prediction of observed choices and functional coupling of other areas of the valuation system and baseline activity was also observed (Engelmann).
On the topic of value and vmPFC, the VS is not the only area where connectivity is important. Unrelated to Kim´s study, it was found that connectivity between the OFC and basal amygdala (BLA) was linked to risk for example, The OFC-BLA pathway has been shown to be important in the decision-making process for appropriate assessment of reward value to influence choice (Zeeb), Rhodes) and whereas BLA lesions cause increased choice of large risky rewards (Orsini), OFC lesions significantly decrease risk-taking (Orsini). Connectivity between the OFC and anterior cingulate gyrus (ACC) is also thought to be important. In this case, neurons in the monkey OFC predominantly encode rewards that are delivered to oneself and neurons in the ACC encode reward allocations to the other monkey, to oneself or both (Chang).
Activities of completely different brain regions are not the only factors where value-based decision-making can be affected. Sub-areas of a region can also be important, but in an indirect way. Kim, Kim and Kim looked at two areas of the medial PFC, ie. ventral medial PFC and dorsomedial PFC (dmPFC) and found different activity patterns which were explained by different functionality in the decision-making task being undertaken. As given above, the vmPFC activity was linked to value, but dmPFC activity was attributed to additional cognitive processing. In this case, Kim found in the observation cue phase that dmPFC and temporoparietal junction (TPJ) activities demonstrated a correlation to social conformity (SC). With high SC, then dmPFC/TPJ activity was greater for the private condition than the public condition with the superior partner than the inferior. With low SC then there was no significant effect for either superior or inferior partners. Kim explained these findings as the greater dmPFC and TPJ activities being involved in additional cognitive processes required for the consideration of other`s opinions plus the observational context which facilitates the detection and resolution of conflicts in the private condition. This explanation is likely since it is known that activity in the dmPFC is modulated to upcoming action value and that activity determines behaviour and not the reward value (Morris). Therefore, it was concluded that the decision is generated in another brain area and then conveyed to the dorsal medial DA neurons which then affect basal ganglia efficiency (Morris). Therefore, with reference to Kim`s decision-making tasks the dmPFC is likely to be a ´relay` area and not the primary working area dealing with value judgements. Also, from the TPJ aspect it is known that association areas are located at the intersection of posterior end of superior temporal sulcus, inferior parietal lobe and lateral occipital cortex. Therefore, information is integrated from the thalamus, limbic system and visual, auditory and somatosensory systems and so information from both external and internal environments can be incorporated. Functionally, the right side is involved in processing of information to orient attention to new stimuli and the left side contains both the Wernicke area and angular gyrus involved in language cognition and processing) and can also play important roles in reasoning of others` beliefs, intentions and desires. Therefore, the observation that activity in both the dmPFC and TPJ by Kim and colleagues in the study participants completion of their value-based decision-making task is not unexpected, but both areas are indirectly linked to value.
In conclusion of the particular direct association at medial PFC activity and value, it should be noted Kim, Kim and Kim could also have looked at a different area of the striatum as well as the ventral striatum. It is known that the dorsolateral striatum is linked to trade-off between computational simplicity and flexible and efficient use of experience (in this case, face values) and this comes about through competition between the dorsolateral striatum and PFC systems (Daw). It is known activity in the dorsal striatum provides motivation to perform in reward-based tasks (Wang). Therefore, it is likely that had Kim, Kim and Kim investigated the vmPFC and dorsolateral striatum area activities would have been seen in the same way as vmPFC and VS and the same observations regarding choice and conclusions would have been drawn.
However, as given above, value is not the only function of the vmPFC relevant to Kim, Kim and Kim´s experimental decision-making task. Activity of this area is also linked to attention which is known to influence the decision-making process. For example, selective attention to relevant information is required for decision-making (Vaidya); the level of awareness is important with not all subconscious events leading to early awareness (Miller); and spatial attention leads to reinforcement learning and requires integration of reward, attention and executive processes (Jarbo). The activity observed in the vmPFC associated with decision-making appears to be important to keeping the focus on relevant options (Noonan) and changes in the 5HT or DA levels in the PFC (particularly vmPFC), striatum and ACC can lead to changes in attention (Settles) plus vmPFC lesions impair learning about relevant material, but does not increase reward attribution to irrelevant material (Vaidya). Therefore, with relevance to Kim, Kim and Kim`s task, activity observed in the vmPFC serves a dual function – value and attention on relevant material required in the decision-making process particularly for the Second Preference Rating.
Another aspect of the decision-making task of Kim, Kim and Kim linked to attention was investigated by looking at another area of the medial PFC that of the rostromedial PFC (rmPFC). In this case, Kim and colleagues found that rmPFC activity in connection with SC and found that in cases of high SC then greater activity was observed when the decisions were congruent with the superior partner than when incongruent. When the SC was low then activity was greater when the decision was incongruent for the inferior partner than congruent. Kim, Kim and Kim said that this showed that rmPFC activity was involved in additional cognitive processes when considering others opinions and either agreement of disagreement with them. This explanation is supported by the known roles of the rmPFC that of attention in multitasking situations, eg. in classical eyeblink conditioning in rabbits (Caro-Martin) demonstrating roles for the area in working memory and time-based prospective memory (Ramnani). This area appears to be required for multitasking (Volle) with the medial rostral PFC area playing a role in the maintenance of attention on external stimuli whereas the lateral rPFC is associated with attention on internal cognitions (Kreplin). It is also known that introspective ability in confidence relating to accuracy of decisions is predicted by the measure of functional connectivity between the vmPFC and rostrolateral PFC (De Martino). Therefore, again the successful performance of Kim, Kim and Kim`s experimental value-based task would demand attentional capabilities and so activity of the rostromedial PFC would be expected.
Therefore, successful performance of Kim, Kim and Kim´s value-based decision-making task led to expected activities in medial PFC brain areas linked to direct functionality of value and indirect functionality of attention being observed. However, this is where the problem of all experimental set-up relating to this type of task lies. Brain areas often have multiple functionality dependent on what task is being undertaken and so it is difficult to ascertain whether the specific task has caused the activity or the general need for activity in that particular area to provide that cognitive capability demanded or a by-product of the task initiation or response. Again, only specific controls will aid in the separation of the two.
Since we`re talking about the topic……………
………….increasing age is found to cause less optimal decisions being made (Goh). If Kim and team`s experiments were repeated but with older participants can we assume that the level of changes allowed in the Second Preference Ratings phase would be lower as participants desire to conform to the superior partner`s opinions decrease?
……….cognitive training is reported to have no effect on value-based decision-making of this type. Therefore, if the participants were given specific training in order to improve choice of strategy and confidence would we expect to see changes in level of social conformity and longer reaction times as options are weighed if Kim`s experiments were repeated?
………administration of certain drugs is reported to change levels of risk, eg. ecstasy (Morgan). If participants of Kim`s experiments were pre-administered with ecstasy for example and the experiments repeated would findings show changes in Second Preference Ratings remonstrative to the lower desire to conform to the superior partner`s opinion as dictated by ecstasy increasing the level of risk?
……background anxiety is reported to change decisions (Engelmann). Can we assume that increasing levels of anxiety from reasons unrelated to the future task would still cause changes in both First and Second Preference Ratings if Kim`s experiments were repeated?
neurochitchat 