Prior research has found that the ventral striatum and ventral prefrontal cortex are very involved in tasks that arouse emotion and require decisions about valued stimuli. Some fMRI studies have reported that the ventral striatum, a dopamine-innervated region of the basal gangia (discussed in a prior post), is more active when subjects perceive an error in their reward expectations such as a sudden lottery win. Other work suggests that the same region instead responds to the value of stimuli on the screen, such as how much you like a product. Unfortunately, most previous experiements cannot dissociate value from prediction error. This frames the basic question in Hare et al: attempt to construct a task that can dissociate these two important components of people’s reactions to stimuli. Also, they aim to support their prior research showing that ventral regions of the prefrontal cortex are the source for value coding.

The task is a bit complicated to understand at first, so try to bear with the long explanation. The basic setup is that you (the study subject) are placed in the fMRI scanner after fasting for a while. In the scanner you are shopping for snack foods that you might be able to actually eat afterwards. Snacks come up along with a price displayed above them, and you decide to buy or not buy the snack at that price. Out of 300 choices over 50 snacks, one choice will be randomly selected to count for real at the end. This means that you don’t have to worry about buying too much and going over budget.

In this basic task, however, the interesting variables are all highly correlated – how much you value a snack, how surprising the snack and price are, and the difference between how much you like the snack and its price – and this poses intractable problems for regression analysis of the imaging data.

But Hare et al. are really clever in finding a way to dissociate these variables. First, along with the price of the snack (which you would have to pay if you bought it and then if that trial was picked to count), there is a random gain or loss on each trial. You get this gain or less even if you decide not to buy the product. For example, you could say ‘No’ to a $3 Snickers, but if there was a ‘+3′ displayed below the snack, you’ll still collect $3 (and if it was a ‘-3′ would lose you $3 independent of your choice).

Then, they also include trials where some part of this setup – the product, the price, and the random gain – are missing. If the snack is present without a price or random gain, you could buy it for nothing (13% of trials). On trials where the snack is missing (and replaced by a yellow square), there might still be a random gain or loss that you get on that trial (13% of trials). Or, on the somewhat odd 13% of trials where there was only a price present above the yellow square, you might be choosing to pay $2 it.

This design is an attempt to dissociate the neural response to three key variables: prediction error, goal value, and decision value.

1. Prediction error: At the onset of a product, there is a prediction error, because you didn’t expect the next event to be more than further fixation, or for the value of the next thing to differ from the average value of all the snacks you’ve already seen. But perhaps you like the presented snack a lot, so the difference between your low expectations and the current snack is high. This is a positive prediction error. Similarly, if you don’t like the snack but you expected an average value snack, the prediction error is negative. But prediction errors don’t actually go into the proposed decision calculation – all you need for that are the next two variables. Prediction errors as used here originate in machine learning algorithms developed in the 80′s, in particular temporal difference (TD) learning. The idea is that a robot (or organism) can function well in the environment if it is constantly making predictions about the value of it’s next state and the difference between what it experiences and its prediction.

2. Goal value: When a snack appears on the screen, something in the brain probably codes for how much you value that snack. The authors used pre-task bids for each of the snacks to compute a correlate with bid value and neural activity on a given trial. Goal value doesn’t depend on the random gain or the price, so a cleaner assessment of goal value is achieved on the trials where a snack is presented alone on the screen.

3. Decision value: When you see a snack you like, but the price is higher than you’d bid, you’ll probably decide you don’t want the item. That’s an example of a negative decision value: your willingness-to-pay (goal value) is less than the offered price. On the other hand, if the price displayed for a Snickers is $1 and you’d pay $4, then the decision value is the positive difference, $3.

The novel task design, with random gains and some trials missing components, greatly decreases the correlation of variables and should allow for these variables to be modeled separately (in technical terms, putting highly correlated variables into an fMRI analysis results in a huge mess).  Prediction error and goal values are now only correlated at r=0.30 instead of r=0.78, and the correlation between goal values and decision values drops to r=0.57 from r=0.85.

Using these three regressors in the imaging analysis, the authors find that activity (BOLD response) in the ventral striatum is correlated with prediction errors, that activity in the medial orbitofrontal cortex (mOFC) is correlated with goal values, and that activity in the central orbitofrontal cortex (cOFC; right above the eyeballs) is correlated with decision values. These results are supported by further analyses using different orthogonalization (basically, de-correlation) of regressors and by the display of timecourses for high and low values of the three primary variables.

From these findings, the authors argue that the results of a large number of prior studies have incorrectly concluded that the ventral striatum codes for goal values (which is actually often not the original papers’ choice of terminology) and that the prior results in the striatum are better interpreted in terms of prediction errors. They note in closing that the localization and dissociation offered by these results can inform future studies of decision making in brain-damaged patients and disease. Next up: critique and discussion.

Emotion regulation is a topic with many unanswered questions. How do we (and when should we) alter the intensity or the nature of our emotions? Which populations are most and least adept at emotion regulation? And inevitably, which neural circuits support regulation?

Research on these questions up to this point has tended to study emotion regulation in isolation (Oschner & Gross, 2005). Subjects typically view a stream of emotionally laden stimuli and modify their emotional responses according to the experimenter’s instructions. Common sense suggests that real-life emotion regulation is intermingled with all manner of other cognitions and behaviors. While taking an exam, we might pause to suppress feelings of anxiety before going on. During social interactions, we might curb emotional impulses—like snapping at an irritating colleague—in the midst of conversation. Surprisingly, relatively little research has investigated the effects of emotion regulation on subsequent behavior.

Deveney & Pizzagalli (2008) buck this trend. They tested the effects of emotion regulation on the processing on a subsequently presented word. On each trial, subjects viewed a negatively or neutrally valenced image, taken, as usual, from the IAPS set). Next, they enhanced, suppressed, or maintained the emotion elicited by the image, according to the experimenter’s instructions. Last, they read a word and decided whether it was negative (e.g. “frustrated”, “whore”, “syphilis”) or neutral (e.g. “hairpin”, “hammer”, “haphazard”). The authors report that up-regulation of negative affect attenuated the P300 component of the event-related potential (ERP) evoked by word stimuli. In addition, down-regulation decreased the word-evoked N400 component.

Before getting into their results, it may be helpful to review some basics of ERP. ERPs are derived from the electroencephalogram (EEG), a measure of brain electrical activity recorded at the scalp. To compute ERPs, segments of EEG activity time-locked to events of interest (in the case of the present study, the presentation of negative and neutral words) are extracted and averaged together. This procedure yields a measurement of brain activity specifically related to a particular type of event. ERP waveforms typically consist of a series of peaks and valleys (see Fig. 3, panels A & B). These local maxima and minima are termed “components”. These components are often referred to by their polarity and approximate latency. Thus, the P300 is a positive-polarity component that occurs about 300ms after the relevant event (for a fuller review of this method, and of common ERP components, see Fabiani, Gratton, & Federmeier, 2007).

The N400 component is most closely associated with the processing of semantic incongruity. First reported by Kutas & Hillyard (1980), the N400 is evoked by words that occur out of context. In the sentence “The graduate student lived in a tenement”, the word “tenement” would not elicit an N400 because it makes sense in the context of the sentence. In the sentence “The graduate student lived in a penthouse”, the word “penthouse” would elicit an N400 because it clashes with the semantic context.

The P300 is an ERP component elicited by task-relevant and attention-grabbing stimuli. Theories of the P300 have related it to updating of task context and allocation of cognitive resources. Although over 40 years have passed since the first report of the P300 (Sutton et al., 1965), the psychological processes reflected by this component remain uncertain.

Deveney & Pizzagalli report that negative words elicit a smaller N400 than neutral words. This makes sense because these words appear in the “context” of a foregoing negative IAPS slide, negative words should be less incongruous than neutral words. This isn’t a novel result—the authors cite several studies with comparable results. Their real contribution here is the effect of prior regulatory strategy on N400 amplitude. Suppression of negative IAPS images decreased the N400 elicited by subsequent words, regardless of whether those words were negative or neutral. The authors argue that suppressing a negative emotion reduces the incongruity elicited by a subsequently presented word.

Similarly, the authors report main effects of word valence and emotion regulation instruction on P300 amplitude. Negative words evoked higher-amplitude P300 than neutral words. At the same time, willful enhancement of foregoing negative affect decreased P300 amplitude relative to “suppress” and “maintain” emotion regulation instructions. The authors interpret these results in terms of resource-allocation theories of P300: since negative emotions are thought to sap resources, up-regulation of these emotions should tamp down the P300.

This study has much to recommend it. The authors approach the topic with creativity, addressing a neglected aspect of emotion regulation. Their design and analysis are straightforward, and their results are statistically robust. Last, their N400 results are clear and compelling. However, interpretive difficulties cloud their results with respect to the P300.

The authors contend that enhancement of negative affect consumes cognitive resources, diminishing subsequent P300 amplitude. However, this claim does not jibe with the main effect of word valence on P300 reported here: negative words evoked larger P300 than neutral words. The authors also cite several studies showing that negative affect attenuates the P300. Inconsistencies both within the present study and between this and other studies complicate interpretation of the present P300 results.

I am troubled less by this minor ambiguity than by the question of the overall significance of this research. What, after all, can the authors claim to have demonstrated? As far as I can tell, their claim is simply that enhancement and suppression of negative affect, respectively, increase and decrease negative emotional responses to IAPS images. This claim is tantamount to tautology. For example, suppressing a negative emotion should make that emotion less intense, decreasing the semantic difference between that emotion and a later less-valenced word. This is perfectly reasonable, but it doesn’t posit anything special about emotion regulation. In this sense, the present results serve only to demonstrate that emotion regulation works—and they muster very indirect evidence to do so.

The authors hint that emotion regulation may do something other than changing levels of emotion. Specifically, they suggest that regulation may itself be a demanding process that has consequences for cognition independent of the particular emotions regulated or the nature of the regulatory process in play. In my view, this is the most interesting thread of the paper. Future studies of emotion regulation should try to disentangle modulatory effects (effects mediated by changes in emotion intensity) from direct effects (those closely linked to the regulatory process), and should test the consequences of both kinds of effects on cognition.

Three nice things about this paper:

1) It contains a nice library of JDM biases
2) Impressive data: 7 studies showing a lack of correlation between SAT scores and propensity for bias (though imo, the anchoring data is ambivalent), then something like 7 more showing a relationship
3) I don’t usually look forward to reading discussion sections, but the authors do a nice job of bringing everything together into a framework to predict when bias will occur. Whether or not the model is correct, it breaks down the judgment process into meaningful units.

I’d say overall that the paper was thought-provoking whether or not you buy everything, and one of the nicest in the general domain of dual-process reasoning that I’ve read in a while.

Throughout this study, the participants are presented with two pictures sequentially. Depending on task requirements, they either need to memorize the brightness of the first image or their feelings towards the first image. Their task is always to report whether the second image is more or less intense than the first along the encoded dimension. The authors categorized trials based on this dimension (brightness or affect) and whether the two images were near or far from each other on a normative scale. Below are some of my thoughts on the study’s methodology.  [For a more detailed explanation of the design, see Amitai's introduction]

Experiment 1 tests the validity of the maintenance task. The authors demonstrated that the participants are more better able to make affective or brightness judgment when the two stimuli are more different on affective or brightness domain. This pattern of results suggests that this design is sensitive in examining the amount of information maintained.  In Experiment 2, they applied different types of distraction tasks during the maintenance period and examined how they affect maintenance of affective & brightness information. However, their distraction tasks are not ideal and don’t seem to interference with the task too much (less than 5% reduction in accuracy) In fact, participants seem to be much better (10% better for “near” comparison) when there are distraction tasks (Exp 2) than when there is no distraction task (Exp 1) (This could be due to group differences)
1.    Counting task (to disable the phonological loop): is only applied when the participants are encoding the first image. This is problematic as the participants can maintain affective information verbally.
2.    Visual search task (to disable the visuospatial sketchpad): the average reaction times are not shown in this paper but it is unlikely that participants need 6 seconds (the length of the search trial) to complete the task. It’s a fairly easy visual search task and should require participants no more than 3 seconds to perform the task. If this is true, the visuospatial sketchpad is also likely to be available for the maintenance of affective information. Hence, it is possible for the participants to store their feelings towards a picture in a visual format and not be influenced by the concurrent visual search task.
3.    Affective regulation task (to disable the affective maintenance system): The participants are required to reevaluate the intensity of a picture. It might be the similarity between the stimuli (the one they need to remember and the one they need to regulate) that causes the confusion. Consequently, worse memory performance was observed. Alternatively, it is possible for the participants to use a verbal code to memorize the affective information and the affective regulation task requires verbal involvement (I find it hard to reevaluate a stimulus without verbal involvement but also have difficulty finding a study that examines this hypothesis). As a result, the participants showed poorer performance for maintenance of affective information.

In sum, it’s not very surprising that affective maintenance is not affected by the counting & visual search task. The authors were amazed by the fact that the affective regulation task impaired the maintenance of affective information. I am, on the other hand, reluctant to accept their interpretation of the results. I am more inclined to believe that affective regulation impairs maintenance of affective information because they both rely on verbal processing.

In Experiment 3, the authors set out to test whether such maintenance is negatively biased. They cited a study conducted by Kensinger & Corkin (2003) which showed that maintenance of negatively valenced verbal stimuli does not significantly differ from that of neutral ones. Mikels et al. (2008) made a very weird hypothesis that if maintenance of affective information used similar means of maintenance (using a phonological loop), no negative bias would emerge. Since the effect of the visuospatial sketchpad is not examined, one doesn’t necessarily have to propose a new domain for maintenance of affective information even when negative bias has emerged. The authors also fail to acknowledge the fact that Kensinger & Corkin did observe significantly slower reaction times for the maintenance of negatively valenced stimuli. Mikels et al. (2008) also underestimate the number of other factors that might have led to the part of the null results found in Kensinger & Corkin’s study (e.g., the intensity of the stimuli used). For example, in Kensinger & Corkin’s study, negative faces (more emotionally intense) are found to have a greater influence on working memory than negative words.

In Experiment 4, the authors presented two pictures simultaneously, and asked the participants to perform a similar brightness or affective comparison task. They claim that this design remove the maintenance component which is present in the first three experiments and whatever difference observed should be due to this reason alone. However, by presenting two pictures simultaneously it affects the way each picture might be evaluated than when presented sequentially. It might be easier for people to distinguish a very happy compared to a slightly happy picture when they are presented simultaneously versus sequentially. However, such difference in stimulus presentation might not affect the evaluation for negative images.

To conclude, it’s an interesting idea to test if a specific system is needed for the maintenance of affective information. Unfortunately, this article fails to provide convincing evidence to support such a hypothesis. A more optimal design is needed to test if we really need a specific system for the maintenance of affective information. For example, one might want to use a distracting task that is truly distracting. Also, one might want to design a task in which it is less likely for people to memorize their feelings verbally. One might also want to consider if there is really a need for this affective maintenance process when other existing models already provide a solution to the maintenance of affective information without proposing a unique component to do so (e.g., Phil Barnard’s Interacting Cognitive Subsystem model www.mrc-cbu.cam.ac.uk/people/phil.barnard/).

Kensinger, E., & Corkin, S. (2003). Effects of negative emotional content on working memory and long-term memory. Emotion (3), 378-393

Watanabe, M., Hikosaka, K., Sakagami, M., & Shirakawa, S. (2007). Reward expectancy-related prefrontal neuronal activities: Are they neural substrates of “affective” working memory? Cortex (43), 53-64.

Anyone who learned about memory even briefly in an Intro Psych course (since the 70’s) is likely to have heard about the concept of a “working memory,” a system that holds things in our minds for a brief period of time. When Baddeley & Hitch proposed the model it came ready with a storage buffer for verbal information (the phonological loop) and for graphic information (the visuospatial sketchpad). One of the great experimental thrusts of these proposed separable components has been to allow researchers to selectively interfere with one buffer or the other with an appropriate task, and see whether one form of interference selectively impairs performance on the task of interest (suggesting a role for that component in task performance).

While Baddeley has since proposed adding an ‘episodic buffer’ to the model, Mikels et al. (2008) argue that more attention ought to be paid to the question of whether and/or how it is that we hold a feeling in memory over brief periods. In the paper reviewed below, Mikels and colleagues suggest that there may in fact be a separate component of working memory dedicated to affective information, and they provide putative evidence for this being the case. I will briefly introduce their methods and results, followed by Yang-Ming’s critical analysis thereof.

The authors performed 4 experiments using the same basic task. Images were taken from the IAPS image set and separately piloted for ratings of subjective intensity on scales of affect/feeling and brightness. For the maintenance task itself, each trial involved the presentation of two images in sequence, separated by a retention period, and subjects were instructed either
a) to judge whether the 2nd image was higher or lower intensity of brightness (only affectively neutral images were used for this task) or
b) to judge whether the feelings evoked by the 2nd image were more or less intense than those evoked by the 1st one.

The authors scored these judgments relative to the pilot (“normative”) ratings, and split them by difficulty based on how close the 2 images fell in the normative ratings (“far” vs. “near”). Expt 1 confirmed that this worked and that “near” trials were indeed harder than “far” for both tasks. Expt 2 used either a “cognitive” (counting and visual search) or “affective” (emotion regulation) task to try to selectively interfere with performance on the brightness and affective maintenance tasks, and showed that performing emotion regulation on a separate image during the retention period interfered with the affective task but the cognitive interference (i.e. tying up the phonological loop and V-S sketchpad) actually improved performance.

Finally, Expt 3 showed that the image valence was important to maintenance on the affective task, revealing a bias for reporting a 2nd negative image as more intense than a 2nd positive image, which the authors suggest is an indication that the negativity bias seen in episodic memory holds for working memory as well. Expt 4 removed the retention interval from Expt 3 (showed the images side-by-side) and this valence effect was eliminated.

Click here to continue on to Yang-Ming’s response…

Certainly their results do not demonstrate that subjects actively reinterpreted the value of their reward.  It is equally plausible that they were merely distracting themselves, especially given that the instructions explicitly instruct them to do so—they were told to think of how blue can have a different meaning rather than how insignificant $4 is.  Such instructions would seem to encourage disengagement from the task rather than a reappraisal of it.  The extent to which participants used either strategy was not measured, so we have no way of knowing their strategies, not even from subjective reports.

This is an unsatisfying feature of the study because the link between CS and reward has not necessarily been attenuated.  If you’re not thinking of a reward, you’re probably not going to be much influenced by it.  Put another way, just because you’re not afraid when you don’t think of spiders doesn’t mean you don’t find spiders frightening.  As far as we know, all that the authors have shown is that people can successfully redirect their thoughts when asked to do so.  It hardly shows that the reward has become decoupled from the conditioned stimulus as a result of this instruction.

And as Kacey observes, there was no monitoring of whether subjects even had knowledge of the outcome, since the CS was probabilistic, and people in regulation trials could have been closing their eyes or looking away.  It would be nice to verify that processing of the reward occurred at least on some level before making grand claims about how reactions to rewards changed between attend vs. regulate conditions.

A notable property of the experiment is there are no behavioral results reported, only fMRI and SCR (although, bizarrely, they refer to skin conductance as their “behavioral” measure).  One wonders what unique contribution fMRI provides in light of the psychophysiological response.  At most the fMRI confirms what previous studies have already indicated about the striatum (it is a “reward area,” one of many) and miscellaneous frontal regions (almost every part of frontal cortex has been linked at some point to cognitive control; cf. the meta-analysis in Oschner & Gross, 2005).

Delgato &al. acknowledge that actual behavioral changes have not been demonstrated, and I agree with Kacey that if we can show that, then we’ll really have something interesting on our hands, especially if it’s sustainable over the long term.  Several questions must be answered before this emotion regulation paradigm can be said to have any practical or theoretical significance.  How permanent is the observed effect?  How deeply has one’s conception of the CS been altered?  What happens when the mind inevitably wanders back from the source of distraction?  (I am reminded of Roy Baumeister’s work showing that self control is akin to a muscle that can be exhausted from use.)  Can an addict teach herself not to experience arousal at the sight of drug paraphernalia without having to engage the strategy of distraction?  Is there a such thing as self-induced extinction?

For the time being, the Delgado &al. paper seems to be about preventing a reaction to the stimulus before encountering it (attentional regulation) rather than changing an appraisal of the stimulus (cognitive regulation).  Distracting oneself from a potential reward is a common strategy for modifying one’s own undesirable behavior.  But such tactics are not tantamount to a manipulation of values or emotions—rather, they place us in a position such that these values have minimal influence over behavior.

Further reading: See Kacey’s post for an excellent introduction to this paper.

We speak as though it’s obvious that money and emotions are intricately intertwined. If this is the case, why isn’t the Fed paying more attention to our mental health? Why aren’t clinical psychologists promoting financial security along with emotional well-being? Surprisingly, the research fields of economics and psychology have long ignored this connection, and instead it has been the newly burgeoning field of neuroeconomics that has stepped in to bridge the gap.

Recently, scientists and NYU and Rutgers demonstrated an important new link between finances and feelings. A study by Maurico Delgado, Meredith Gillis, and Liz Phelps just published in Nature Neuroscience (http://www.nature.com/neuro/journal/vaop/ncurrent/full/nn.2141.html) showed that both the physiological and neural response elicited in anticipation of a positive monetary reward ($4.00) can be attenuated by cognitive reappraisal.

They used a standard classical conditioning paradigm where subjects learned that a particular colored square predicted the chance to win money, while another colored square predicted no money. This is akin to walking up to two slot machines, one you know might give you a $4.00 payoff and one you know is out of order. Each time you pull handle on the working slot machine, you feel some excitement as the wheels spin, waiting to see if this time might be your lucky break. Pulling the handle on $0 slot machine just doesn’t feel as exciting; you already know the outcome is $0 before you start. Likewise, Delgado, Gillis, and Phelps showed that people’s bodily arousal increased when they thought they might win money, as did neural activity in the ventral striatum, a part of the brain important for predicting positive outcomes such as wining money or drinking juice.

Here comes the cool part. They then asked their subjects to think about the colored squares differently. Rather than focusing on how much money they might win, the participants were told to think about more abstract qualities of the cues, like how a blue square could represent the blue ocean. This type of reappraisal procedure has been widely used in emotion research to show how cognitive regulation can change the physical and mental experience of an emotional reaction. For example, people viewing a gory amputation video can be instructed to reappraise the situation and think of the clip as an instructional medical film, and indeed people show a resulting drop in the bodily, neural, and subjective experience of disgust. Here, Delgado, Gillis, and Phelps showed that emotion regulation techniques work just as effectively on monetary stimuli as they do on typical emotion-inducing films or pictures. Reappraising during anticipation of the potential $4.00 reward resulted in decreased bodily arousal (measured by skin conductance, an index of how much you sweat on a millisecond timescale), and reduced neural activation in reward prediction brain regions (measured by blood oxygenation levels in the brain via fMRI, which serves as a proxy for neural activity).

The present study is (somewhat surprisingly) the first to directly bridge the fields of emotion regulation and reward processing. Not only can a subjective emotional experience be regulated by cognitive reappraisals and reframing, but cold hard cash can also seem less exciting and motivating if you think about it differently. Their finding is both basic and surprisingly provocative at the same time.

Basic because we’ve known for a long time that our experiences of monetary outcomes can be affected by the way they are framed. Standard Economic Utility Theory described diminishing marginal returns, explaining that the same amount of money just doesn’t mean as much if you are starting from a larger base amount. Prospect Theory told us that it makes a difference whether you’re in the red or the black, that losses loom larger than gains. In the terminology of behavioral economics, reappraisal may be just another technique for shifting an individual’s reference point.

Basic also because we’ve known that placebo effects work in a similar way when people are anticipating a painful or aversive stimulus. If you tell a person that you will give them a special analgesic cream before administering a painful burning stimulus or an electric shock, they subjectively report less pain, their bodily response to the noxious stimulus is diminished, and “pain regions” in their brain show less activity. So, it’s not surprising that the anticipation of positive events can be similarly downplayed in the body and the brain.

However, their findings are also fairly provocative. We know of many animals that can be classically conditioned to anticipate a reward (monkeys, rats, sea slugs), but no other animals have yet shown that they can attenuate their own conditioned response just by thinking about it differently. Pavlov’s dog wouldn’t have ceased salivating if you told him to appreciate the beautiful ring of the bell instead of the food immediately following it.

There are a few aspects of the study that leave me curious. I’d like to know *by how much* each subject was able to reappraise their potential reward. For example, does a down-regulated $4.00 reward seem more like $2.00? $1.00? $0.00? Would different people be able to reappraise down to different amounts? (and would this enhanced down-regulation correspond to greater brain activity in “emotion regulation regions”?). You could start to get at this question by using varying dollar amounts in the conditioning procedure.

Furthermore, a person can avoid thinking about the cash reward with a number of different strategies, including closing his eyes and not witnessing the outcome at all. It is unfortunate that the task design can’t ensure that participants were paying attention to the outcomes, as in making a yes/no button press if they won after the outcome period.

Finally, it’s important to clarify that this study is about the anticipation of monetary rewards. Their findings do not say anything about how the experience of actually receiving or not receiving $4.00 is altered by cognitive reappraisal (in fact, they excluded trials where subjects won money from their analyses). Most likely, monetary outcomes would be similarly down-regulated by cognitive reappraisals, but since we know that anticipation and outcome processing are mediated by somewhat different neural circuits, perhaps emotion regulation resources would be differently deployed. Additionally, it’s possible that regulating during anticipation versus outcome would have different behavioral effects, similar to the way reappraisal and suppression differ in emotion regulation. Reappraisal refers to the way a person interprets an emotional event as she is experiencing it, while suppression involves controlling outward reactions like facial expression and heart rate. Though both techniques can be employed to regulate emotion, reappraisal is often more effective and less effortful, perhaps in part because it works earlier in the time course of an emotional reaction.

An obvious follow-up to the current study is to test the hypothesis that emotion regulation strategies can alter decision-making. However, operationalizing value-based decision making is a tricky thing. What kind of decision making paradigm do you use? A decision between risky and safe gambles? Bidding money on food items while hungry? Choosing between a smaller immediate versus a larger delayed reward?
Given the current evidence that anticipatory affect can be down-regulated by cognitive reappraisal, the big question remains: does anticipatory affect = valuation? If so, how do you ensure that reappraisal can selectively diminish an individual’s value of one thing (say, $4.00) without diminishing the value of the thing it is being compared to (say, a candy bar)?

Overall, the study by Delgado, Gillis, and Phelps is simple, elegant, and long-overdue. That our thoughts can intrinsically shape our experience of a basic monetary reward offers an important lesson in the phenomenal power of our own minds.

- Kacey Ballard, Stanford University

Read on: Nina’s take on the Delgado paper.

Don’t forget to send us feedback on reading & presentation requests! We’re trying to divide the labor of reading list development as much as possible, and this also gives us a sense of what other people are interested in.

Note: Signing up does not commit you to anything, in terms of posting responsibilities. It merely gives us a guideline for who to contact from time to time about presenting.