 The broad topic of our research project is beliefs. Now beliefs are of course very important for economics. Think about the decision to buy a house or to invest in the stock market. You're going to have to form beliefs about how the interest rate is going to develop or how the stock market or the housing market is going to develop. Now given the importance of beliefs, it also seems important to understand how beliefs are formed. And so in the broad sense what we try to do here is we try to improve our standing of the process of belief formation, focusing on a specific aspect which is how does the way our memory works actually shape the process of belief formation. Now memory is of course a very broad concept and we focus on a specific aspect of memory which is associativeness. Now this principle of associative recall basically says that the recall of memory trace is triggered by cues that are mentally associated with that trace. And now these associations can be constructed in many different ways but an important determinant here is similarity. What this means is that the recall of a memory trace becomes more likely if this memory trace is similar to the cue that you are currently observing. So the basic research question we ask in this project is how does associative recall shape the process of belief formation and how does associative recall determine to what extent you take current information into account. To make this a little bit more concrete think again about the housing market for example. What we are interested in is suppose you now obtain a piece of information about the housing market but in the past you have already made some experiences or obtained information about the housing market as well. What we are interested in is to what extent as the current piece of news you observe about the housing market shape the recall of the past information you learned about the housing market. So which pieces of information about housing that you already know become accessible because they are triggered by the current piece of information you are reading. So the question we try to answer is how does the receipt of a current piece of information shape the recall of past pieces of information you have obtained. In order to answer this question we conduct a controlled lab experiment and in doing so we were facing in particular one challenge which is that of course we needed an experiment where memory and memory constraints can actually play a role. Now in existing experimental paradigms that try to study belief formation essentially by design memory was playing no role so we had to come up with a new paradigm that allows us to study the role of memory for belief formation. Now the way we do this is as follows the respondents in our experiment had to estimate the value of 12 hypothetical companies. Now for each of these companies our respondents would obtain pieces of news. Some of these news could be positive meaning that the value of the company increased. Some of these pieces of news could be negative meaning that the value of the company actually decreased. After obtaining these pieces of news our subjects would be asked to estimate the value of a given company. Now in terms of timeline there was a first period where subjects could accumulate knowledge about the value of that company. So in these first periods for each company subjects would obtain pieces of news about that value of the company and you can think of these pieces of news as existing information about the value of each company. Now some time was elapsing and then in the second period for each company there was an additional piece of information and now after obtaining this additional piece of information we asked subjects to estimate the value of a given company. Now what this setup allows us to do is to ask exactly the question we raised in the beginning. So to what extent does the last piece of information determine which aspects of the information you obtain in the first period become accessible and again our hypothesis would be that if the last piece of news is positive for example then by the principle of associative recall this would trigger positive pieces of information from period one. So highlight one of our key results we find in our experiment that associative recall causes participants in our study to overreact to the last piece of information they see. What this means is that people seem to put too much weight on this last piece of information so they react too strongly to the last piece of information. If the last piece of information is positive they sort of overvalue the value of that company. If the last piece of news is negative they underestimate the value of that company. Now the intuition for why associative recall can actually generate overreaction is quite straight forward. Think of a respondent who obtained the last piece of information that is actually positive this last piece of news will trigger the recall of past pieces of news that have been positive and so our decision maker will overestimate the value of that company. If the last piece of news our decision maker sees as negative he's going to be more likely to remember the negative pieces of news about that company he has seen in the past and so he's going to underestimate the value of that company. So the beliefs of our respondent will fluctuate too much based on the realization of the last piece of news so he's going to look like he's overreacting to this last piece of news. If you think about that in the context of say the stock market for instance what this means is that it might be that an investor in the stock market observes a piece of news about some development in the stock market and this piece of news is going to trigger the recall of past pieces of news he or she has seen and due to the principle of associativeness this is going to lead him to overreact to this recent piece of news. So I think our findings are relevant first of all because we introduce a new experimental paradigm that actually allows us to study the role of associative recall but also potentially other memory imperfections for belief formation. As I said before in the traditional experimental paradigms essentially by design there was no role or no scope for memory imperfections to play a role. So here we introduce a new experimental environment that allows people to study the role of memory imperfections for belief formation. Second I would say that now there's actually an emerging body of theoretical work that tries to formalize this principle of associative recall and how it affects belief formation and subsequently also behavior. I would say that our findings sort of directly relate to this body of work and directly empirically confirm key assumptions and key mechanisms formalized in these theoretical accounts. Finally our result that associative recall can actually cause overreaction to news nicely relates to another body of existing empirical work that has documented overreaction to news in a diverse set of contexts ranging from the stock market housing market to beliefs about political issues where as I said people have documented overreaction to information but in these contexts it's not really clear where this overreaction is coming from and so one could view our results as sort of potentially providing a micro foundation for why we see overreaction in these contexts. So I think this topic of memory and belief formation is still quite new so there's a wide range of possible extensions or directions for future research. Let me highlight maybe two that are sort of closely related to this project. The first one is the question to what extent this principle of associative recall is actually beneficial for us. So the broader question underlying this is of course to what extent is the way our brain functions beneficial for us. Now for this principle of associative recall of course if you compare it to a situation where our recall is perfect it's not going to be beneficial. But that does seem to be a somewhat unfair comparison because we do know that our memory is not perfect. So maybe the better benchmark or the better comparison would be to a situation where our recall is just random. So we randomly recall or forget stuff and compared to that what our research already shows is that there can be situations where clearly associative recall is better than random recall but there can also be situations where associative recall leads to larger mistakes than if your memory would function in a more random fashion. But much more work needs to be done to really understand to what extent our associative-ness principle in memory is really beneficial. I think another interesting direction for future research would be to look at strategic interactions and the role associative recall can play for such strategic interactions. So for example consider the situation of an interaction between a company and a consumer and the company of course wants to sell a certain product to that consumer. Now this company can potentially exploit this principle of associative recall so the fact that the memory of the consumer will work in an associative fashion. So for instance this company can sort of strategically manipulate the content of their newspaper ads for instance to shape the recall process of potential consumers in a specific way in order to persuade this consumer to buy a certain product. I think sort of better understanding to what extent in such strategic interactions an associative memory can play a role and can be exploited for instance from firms but also from other actors in markets could be an interesting area of future research.