Supervisor: Dr Caroline di Bernardi Luft
We separate the world into categories in order to make sense of it, and to make sense of our own behavior; in the latter case simple categorical performance feedback (Correct vs. Incorrect) and categorical reward (Gains vs. Losses) information both help to shape and maintain behaviour. Performance feedback, in particular, often comes as a more fine-grained signal of our behavior and performance levels (e.g. deviation from target or scores in a game). We are now at a stage that we know reasonably well the neural mechanisms underlying the processing of both categorical performance feedback and rewards and how we learn from those outcomes (Luft, 2014). However, the mechanisms behind the processing of more fine-graded feedback have been vastly overlooked even though it is one of the most usual forms of performance outcome (from driving a car, to getting our grades in exams). The few studies (Fromer, Sturmer, & Sommer, 2016; Luft, Takase, & Bhattacharya, 2014) looking at graded feedback ended up artificially and arbitrarily categorizing the feedback, at the cost of losing one of the most interesting and crucial aspects of it: the subjective evaluation of such quantities. That is, performance outcomes are not clearly good or bad, correct or incorrect, they require an interpretation or a value judgment which can lead to decisions. For example, one student can be very happy with a mark of 7 out of 10 if him or her was under 7 in past exams, whereas another one may consider a mark of 7 as disaster as she had only experienced more than 9 in the past. The first student might decide to take a break and celebrate, whereas the second may decide to focus on studying and prepare better for the next exam.
Therefore, how we interpret graded outcomes is bound to have an impact on our decisions. A crucial question regarding the processing of graded feedback is how these quantitities or values are evaluated and what that this evaluation leads to. Previous studies (Fagerlin, Zikmund-Fisher, & Ubel, 2007; Mishra, Mishra, & Shiv, 2011; Peters et al., 2009) showed that by manipulating the subjective evaluation of finelly graded information (e.g. by providing an “expected average” as a reference to compare each graded outcome to), it is possible change the individual’s perception of an outcome and their actions towards it (e.g. willingness to take a treatment). In this project, we ask how the subjective evaluation of feedback information (e.g. rewards, errors) vary accross individuals and over time. These changes will be investigated not only in relation to individual differences (e.g. arousal levels, motivation), but also over time depending on dynamic revisions of individual representation based on internal and external states (Rangel, Camerer, & Montague, 2008).
This project aims at investigating the dynamics of subjective feedback evalutation. The main questions to be addressed include:
1. What are the neural markers of subjective feedback evaluation and how can we decode them?
2. How does the subjective evaluation of feedback changes over time and during different learning stages?
3. Are there individual differences which can explain subjective evaluation (e.g. personality)?
4. Is it possible to change subjective evaluations by behavioural manipulations and transcranial alternating current brain stimulation (tACS)?
In order to understand the subjective evaluation dynamics, statistical learning and procedural learning tasks will be used with categorical and graded feedback. The neural correlates of the decisions will be investigated using electroencephalography (EEG) on the single-trial level and the analysis will be accomplished by combining advanced signal processing techniques to analysing brain oscillatory responses (both local and global), network analysis (graph theory), and psychophysiological interactions, ECG, and GSR changes over time as well as brain responses to it (e.g. Luft & Bhattacharya, 2015). Supervised machine learning algorithms will be used to decode the subjective evaluations of feedback (Helmstaedter, 2015). Finally, tACS will be used to test the causal role of such oscillations on subjective feedback evaluation. tACS will be combined with EEG in order to investigate its effects on the neural signatures of subjective feedback evaluation.
Interested candidates must email Dr Luft for questions/instructions.