The amount of data created daily and stored globally is measured with zettabytes. One zettabyte equals 2^70 bytes. Digital trackers, in a form of apps, cookies and sensors record our behaviour not only online, but also in a non-digital space when we visit places or have a conversation. Given the large amount of existing and still incoming data about each individual, does it still make sense to ask people about their preferences, opinions, individual traits and experiences? Already in 2013, psychologists Michał Kosiński, David Stillwell and Thore Graepel documented that a user's behaviour in the social media can predict their personality, race, gender and political views. Later extensions of this work showed that other sources of digital footprints, such as browsing history, can be similarly indicative of a person's individuals traits and digital footprints in general can describe a person better than their friends and family. Should these findings make questionnaires and surveys obsolete?

In behavioural sciences, there are two ways of collecting data about individuals: 1) self-reported and 2) behavioural. Self-reported methods, also referred to as psychometric methods, involve directly asking people questions about their preferences, behaviours, experiences and feelings. Behavioural methods measure the same constructs indirectly. Behavioural methods take the form of simple (gamified) tasks or simulations of real-life events to observe how a person behaves. For this reason, behavioural measures are closer to behavioural analytics - applying statistical, econometric, and machine learning analyses to existing data. Many constructs, such as preferences, can be measured using both types of methods. For example, to estimate a person's risk preferences, a researcher could either ask "do you prefer investments with high possible gains and high variance of possible outcomes, or safe bets with low possible gains with low variability of potential earnings?". Alternatively, a researcher could compose a set of lottery pairs reflecting this dilemma and ask a person to choose between a high-risk-high-reward and a low-risk-low-reward lotteries. If the researcher applied these two methods to elicit risk preferences of the same person, would they obtain the same result? It depends, on several things.

First, behavioural methods are highly sensitive to the context and framing of the problem. It is especially the case, when the task involves a numerical problem. Then, a person's numerical abilities - the so-called numeracy can confound or influence their response. Second, behavioural methods may include an insufficient number of observations to obtain the "complete picture" about a person. For example, when estimating a person's preference for dark vs. white chocolate, we may obtain 50/50 (indifference) preference with only four observations, when a person chooses dark chocolate in two cases and white chocolate in two other cases. However, the true preference could be 70% for dark chocolate, where in 7 out of 10 cases, a person would choose dark chocolate over white chocolate. Revealing this true preference would only be possible with a sufficient sample and accounting for external factors like the availability of the favourite brand, and a person's energy levels at the time of sampling.

In contrast, self-reported measures can give a more "complete picture" about a person, within one question. Likely, a person has access to more data about themselves than a limited sampling would. Therefore, asking a question like "How risk taking are you on a scale from 1 to 10" tends to reveal more consistent results than using behavioural measures. At the same time self-reported measures can carry a bias. A person's view of themselves may not converge with their actual behaviour. At the ETH Zurich, we conducted a study in which people exhibited more risky traffic behaviour than how they described themselves. This could be either because of a persons subjective self-view or because of how they want to present themselves to others.

In [1], we created a risk task for eliciting people's risk preferences, which combined behavioural and self-reported features such as context, verbal description and an unbiased selection of choice options. We found that people's consistency in responding to our measure ranked between standard self-reported and behavioural measures. In [2], we further investigated why behavioural measures often provide divergent view of the same person, such that one behavioural risk measure describes a person as a risk taking, while another measure as a risk averse individual. The root cause of obtaining such divergent results was the choice architecture. For example, if a behavioural risk measure offered many low-risk choices, and only a few high-risk choices, people were more likely to choose one of the numerous low-risk choices.

Behavioural measures can be prone to sampling errors and more sensitive to context that cannot be easily captured by behavioural data data. In contrast, self-reported measures may depict what a person thinks about themselves or what they want others to think about them. Using digital footprints is a powerful method of studying people, but it may not give the full picture without asking the right questions. The authors of the study linking social media behaviour with the users' personality measured personality with a version of personality questionnaire and tested whether behavioural data can predict the self-report. Psychometrics is not obsolete in the age of Big Data and asking questions is an essential and a non-trivial task.

Terms to remember: digital footprints, self-reported methods. behavioural methods, behavioural analytics, context and framing, numeracy, choice architecture, sampling error

Publications:

1. Heinke, S., Schürmann, O., Andraszewicz, S. & Rieskamp, J. (2024) Improving Behavioral Risk-Preference Measures:  Many Decisions with Gains and Losses Increase Test--Retest Reliability. Available at SSRN: https://ssrn.com/abstract=4885566 or http://dx.doi.org/10.2139/ssrn.4885566

2. Mechera-Ostrovsky, T., Heinke, S., Andraszewicz, S. & Rieskamp, J. (2022). Cognitive abilities affect decision errors but not risk preferences: A meta-analysis. Psychonomic Bulletin & Review, 29(5), 1719-1750

Originally published on LinkedIn on 24.03.2025

Financial bubbles and crashes have been fascinating academics and finance professionals for years. They can have a big impact on the economy and the wellbeing of societies. At the same time, they are driven by the psychology of many interacting individuals. Since 1980s, behavioural scientists have been using computerised tasks to create artificial financial markets to investigate factors that drive excessive stock prices. This seminal methodology was honoured with The Nobel Prize in economics awarded to Vernon Smith in 2002. Consistent overpricing, followed by a market crash, the so called the "bubble-and-crash" scenario, has been a common finding in the experiments using this methodology, despite variations of the experimental features.

Computer technology has substantially evolved since the 1980s. This has allowed for creating experimental asset markets characterised by high complexity mimicking the real stock markets. Using the xYotta software (https://innoview.ethz.ch/project/xyotta/) developed at the ETH Zürich and Future Resilient Systems (FRS) , we created a market that replicated the trading rules of the Swiss Stock Exchange market (SIX). Over a hundred of semi-experienced investors traded multiple financial assets (up to 40) over a period of several weeks. The traders could place orders at any time and from any place as long as they had internet access. Before we opened the market, we first elicited the investors' belief about the future success of these assets. The success of the assets would depend on an actual real-life event. In real markets an event such as a result of political elections could determine a success of some businesses. After our market closed, we again elicited traders' belief about the success of the assets.

We observed that the market prices mapped the averaged belief of all market participants. The pre- and post-trading belief distribution did not differ much from the asset prices. The price distributions substantially shifted when we introduced "news", which was vague and uncertain information distributed to half randomly selected market participants. We also found that market participants have a tendency for forming a bell-shaped belief distribution, with the middle values being most likely, and extreme values receiving lower probabilities (so called tails). This tendency was surprising in markets, in which historical market data would suggest a uniform distribution of prices. In a uniform distribution, all values are equally likely. We also asked participants to provide their reason for putting an order. About half of the orders were motivated by the traders' "gut feeling".

Next, we replicated this naturalistic experiment in a behavioural laboratory. As before, we found a convergence between the belief and price distributions, despite the fact that the market dynamics substantially differed between the naturalistic and the laboratory settings. These findings raise questions about the importance, origin and aggregation of intuition in decisions in uncertain and complex situations such as financial markets. These decisions have a major impact on the financial well-being of individuals and societies. You can find more details about this series of studies in [1] and [2].

Terms to remember: experimental asset market, bubble-and-crash scenario, gut feeling, belief elicitation, bell-shaped distribution, uniform distribution, naturalistic experiment, behavioural laboratory, intuition

References:

1. Sornette, D., Andraszewicz, S., Wu, K., Murphy, R.O., Rindler, P. & Sanadgol, D. (2020). Overpricing persistence in experimental asset markets with intrinsic uncertainty. Economics: The Open-Access, Open-Assessment E-Journal, 14(20), 1-53, https://doi.org/10.5018/economics-ejournal.ja.2020-20

2. Andraszewicz, S., Wu, K. & Sornette, D. (2020). Behavioural effects and market dynamics in field and laboratory experimental asset markets. Entropy, 22(10), 1183, https://doi.org/10.3390/e22101183

Originally published on LinkedIn on 11.11.2024

Autonomous vehicles are an exciting technology that could revolutionise the way we commute and travel. From March 1st, 2025, Switzerland will allow using certified car autopilots on highways. Drivers will not be required to monitor their self-driving vehicle, but they need to be in a “driving-readiness condition”. Driverless parking in designated areas will also be allowed [1]. Switzerland is not the only country that allows use and testing autonomous vehicles, with certain restrictions. However, moving around busy streets with pedestrians, cyclists and human drivers still poses a challenge to self-driving cars. Past accidents involving first versions of autonomous cars raised concerns whether and how these vehicles should be allowed on streets shared with pedestrians and cyclists. Could behavioural science be used to anticipate pedestrian behaviour in face of self-driving cars?

We tested whether Virtual Reality (VR) can prove useful for investing human interaction with autonomous vehicles. Virtual Reality - an immersive technology that enables simulating various hypothetical situations in a form of computerised games - can help pre-test hypothetical situations before they are introduced in the real life. This approach has been used in architecture to test the functionality of buildings before they are built. We created a street crossing task, in which we  presented people a movie of a car approacing with a constant speed of 30 miles/h. We asked the participants to indicate the last safe moment to cross the street. Some of the cars were labelled as autonomous, some as driven by a human. We presented a movie of a neutral-looking black passenger car with VR and a movie or a real car driving on a street. The angle of the view and the car speed were the same in all cases. We also tested participants’ knowledge about traffic safety and rules- We found no differences in the response time between autonomous and human-driven cars presented in the VR movies. This means that people reacted the same, independently of whether they knew that an approaching car was autonomous or driven by a human. This speaks against technology-aversion theories that would assume higher caution when facing an autonomous car. The most surprising finding of our study was that most of the study participants would enter the street too late, potentially causing an accident. At the same time, most participants correctly responded to the theoretical question about the safe distance to cross the street in front of an approaching car. This represents the gap between intention and behaviour. Also, our respondents were much more cautious when they observed a movie of a real car, rather than any of the cars in the virtual reality setting. When facing a recording of a real car, study participants would enter the street earlier, than in the VR recording. The potential explanation is that VR does not present the spacial dimensions in a sufficiently realistic fashion to let people judge the safe distance from and speed of a moving object. However, various features of the decision environment, such as the physical environment, may also play a role.

The study is an example of the limitations of using immersive technologies for pre-testing human interaction with new risky technologies. It also highlights the complexity of predicting human behaviour, even in simple every-day tasks, like crossing the street. This sparks a discussion on how to design simulators and what applications of virtual and augmented reality are most suitable for. Should we let autonomous vehicles learn to predict how humans would behave, or should we let humans learn how to interact with those vehicles?

Terms to remember: virtual reality (VR), immersive technology, simulator, gap between intention and behaviour, decision environment, technology aversion

References:

[1] Der Bundesrat, Bern 16.12.2024, https://www.admin.ch/gov/de/start/dokumentation/medienmitteilungen.msg-id-103600.html

Originally published on LinkedIn on 23.12.2024

Setting financial goals, such as making savings and investing requires consistency and persistence in applying decision strategies. At the same time, planning your December finances already in January may require some flexibility because the future events are unknown. You may aim for putting a certain amount of money aside, but an unexpected expense may occur forcing you to adjust your goal. Alternatively, you may get an additional gig and obtain enough savings already in the middle of the year. In a similar way, an entrepreneur may consider investing in a new business venture, sequentially putting money in different phases of their business project. The success of these investments is usually unknown until the venture is fully launched.

Sequential investment or regular saving are examples of sequential decision making, in which several decisions lead to a certain outcome. Sequential decision making is prevalent in individual, managerial and entrepreneurial decisions. Investing in a risky asset or a new venture is an example of a dynamic decision making because the outcome of a sequence of investments depends on the evolution of events, until the uncertainty about the success of these investments resolves over time. How well do people navigate in these situations?

To investigate this, we conducted an experiment in which we asked 42 participants to make six sequential investment decisions with available 36 dollars (see [1] for details). At each step, they could invest between 0 and 6 dollars. At each step, we would roll a fair die and count the number of dots that comes out at each roll. If the sum of all six die rolls was at least 24 points, the amount that our participants invested would be doubled and paid out to them. If the sum of the six rolls was less than 24, the participants would lose the invested amount. The optimal solution to this sequential investment task was to invest 0 when the probability of winning was less than 50% and invest the maximum amount of 6 dollars when the probability of winning was at least 50%. In that logic, the optimal strategy would be to invest nothing as long as the sum of the dots from die rolls was less than 12 and invest 6 dollars when the sum of the dots was at least 12. This task requires both, persistent application of the same strategy and changing the strategy when the situation changes.

Naïve diversification, in which participants invest middle amounts of 2 or 3 dollars was prevalent, especially in the early steps of the task. Naïve diversification is a simple heuristic (i.e., mental shortcut) that allows for equal risk distribution (i.e., “risking half”) but it does not profit from the flexibility to maximise gains. We repeated the experiment with a similar sample of 41 participants, to test whether limiting the available funds would reduce this bias. With only 24 dollars of endowment, participants more frequently invested nothing across different steps of the task, but the bias of investing something in the early steps remained. Also, naïve diversification decreased in later steps, indicating that our participants were inconsistent at applying the same strategy.

Making financial resolutions requires both, sticking to your desired startegy and changing it, when needed. Some people in our experiment managed to invest optimally, maximising their gains. However, a large proportion of the participants neither persistently applied the same strategy, nor they adjusted their strategy as the uncertainty about success of their sequential investments resolved. Which of the two groups do you think you belong to?

Terms to remember: dynamic decision making, real options, naïve diversification, sequential investment task

References:

[1] Murphy, R.O., Andraszewicz, S. & Knaus, S.D. (2016). Real options in the laboratory: An experimental study of sequential investment decision. Journal of Behavioural and Experimental Finance, 12, 23-39

Originally published on LinkedIn on 06.01.2025

Social influence is a change in a person's thoughts, feelings, attitudes, or behaviours caused by other people. Such a change can happen through pure observation of others, without the influencers explicitly aiming to influence. Social comparison is an example of this phenomenon, in which a person compares themselves to their peers. Maybe you have ever taken an exam or participated in a race, after which you compared your score with the scores of your peers. If you did, you may have checked how well you did compared to the best person and to those performing worse than you. Both of these mechanisms are very important for our development and well-being. Downward social comparison, such as comparing yourself to worse people within your peer group, helps you maintain your ego and self-esteem. In contrast, upward social comparison, such as looking at the better performers, is a trigger for the goal-directed behaviour. This means that when you compare yourself to someone who over-performs you, you may take actions to catch-up with this person. However, when your efforts do not bring the desired effect or the gap is too large you may experience negative feelings such as frustration.

How does this translate to economic behaviour? In [1], we found that that "keeping with the Joneses" on widely popular trading apps results in people's increased activity and financial risk-taking within the platform, without them obtaining earnings as high as the best performers. Not being able to catch-up with the Joneses was related with people's decreased satisfaction from their earnings. We also found that the increased effort behaviours was a combination of emotionally driven (affective) and deliberately planned (cognitive) drivers of behaviour. These behaviours are controlled by the pre-frontal cortex (i.e., the brain part at the forehead), subcortical brain structures (i.e., parts of the brain located underneath the outermost layer of neurones) that are also part of the limbic brain system regulating emotions, motivation and goal-directed behaviours. These brain structures are also responsible for experiencing reward from our actions. In [2], we found that these brain structures respond differently to rewards in the social and private contexts. In the social context, our participants' brains released the reward signal when the participants performed economically at least as well their peers, while in the private condition their brains released the reward signal when their economic performance was high but did not reach environmentally unsustainable levels.

In both experiments, in [1] and [2], we induced social influence through presenting our participants with information about the performance of their peers who participated in the earlier rounds of the experiments. Therefore, the presented peer performance was pre-recorded and the "influencers" were not informed that their task performance would be presented to other experimental participants. We did not instruct any of the participants to copy or follow others. Social influence can occur through a mere observation of others and our brain is sensitive to this.

Terms to remember: social influence, downward/upward social comparison, goal-directed behaviour, affective and cognitive drivers of behaviour, subcortical brain structures, limbic system, pre-frontal cortex

References:

1. Andraszewicz, S., Kaszás, D., Zeisberger, S. & Hölscher (2023). The influence of upward social comparison on retail trading behaviour. Scientific Reports, 13, 22713, https://doi.org/10.1038/s41598-023-49648-3

2. Martinez-Saito, M., Andraszewicz, S., Klucharev, V. & Rieskamp, J. (2022). Mine or ours? Neural basis of the exploitation of common-pool resources. Social Cognitive and Affective Neuroscience, 19(9), 837-849, https://doi.org/10.1093/scan/nsac008

Originally published on LinkedIn on 09.12.2024

Digitisation has significantly changed the way people deal with their personal finances, including everyday tasks such as paying for goods and services, splitting bills with friends, saving and investing. Digitisation lead to decreased costs of these services, making them easier and available at any time without the need of going to any physical place. This leads to democratisation of financial services. For example, now, it is possible to use an app to buy an fraction of a stock and investing in already diversified products such as ETFs (exchange-traded funds), without the need to possess large amounts of disposable assets and engaging an a professional asset manager. Also, it is not necessary to ask for a change because digital payments allow for paying the exact amount to a single cent. At the same time, digitisation creates space for new products and services that affect users and customers in new ways.

We looked at users' behaviour on social trading platforms - online services for investing in financial assets that allow for connecting and following other users. Social trading platforms  integrate the features of a trading app and social media. Often, they put the best performing users in the spotlight. To other users, this may have a similar psychological effect to watching celebrities on social media. In an experiment simulating a social trading platform (see [1] for details), we found that comparing oneself to the unreachable top performers resulted in investing more in stocks and doing more high-volume and high-value transactions. This increased effort did not result in higher earnings, but it decreased investors' satisfaction from their trading activity. This an example of the effect of upward social comparison on financial behaviour and satisfaction of retail investors (i.e., non-professional investors).

Not only investing, but also payment means move to the digital space. Within a project (see [2]) investigating the links between use of financial technologies (FinTech) and resilience of individuals persons, we found that users of e-wallets have stronger social networks, have better financial knowledge, have better access to financial serviced in general, and are less prone to suffer a mental illness. Using mobile payment technologies is linked to higher social, financial and psychological resilience of individuals. In the same project, we surveyed users of a major crypto-trading platform in Europe to test how risks and opportunities of using cryptocurrencies relate to the users' individual traits. In our study, users of cryptocurrencies demonstrated a fair understanding of basic financial and blockchain-related concepts.

Investing, paying and saving has become easy and accessible for everyone with access to the Internet. Digitisation of personal finance is a process that carries some risks that need to balanced out by the new opportunities. A careful inspection of the risks and a cost-benefit analysis of finance digitalisation should be especially relevant to financial regulators.

Terms to remember: democratisation of financial services, ETF, social trading platforms, a trading app, upward social comparison, a retail investor, an e-wallet, social, financial and psychological resilience

References:

1. Andraszewicz, S., Kaszás, D., Zeisberger, S. & Hölscher (2023). The influence of upward social comparison on retail trading behaviour. Scientific Reports, 13, 22713, https://doi.org/10.1038/s41598-023-49648-3

2. Andraszewicz, S., Roberts, A. C., Wettstein, L., Popelka, D., & Hoelscher, C. (2024). The Resilience and Tech Database: Cross-cultural Datasets Linking Psychological and Financial Resilience with Financial Technology Adoption. https://doi.org/10.31219/osf.io/z8csh

Originally published on LinkedIn on 25.11.2024

Social media apps are on the rise. It is estimated that in February 2025, two thirds of the global population are using social media. The social component has become an integral part of many services, such as online shops, education tools, sporting and health platforms or trading apps. User engagement through social media has become an integral part of marketing, but relying on external platforms poses an additional risk to a business. Why not integrating a social media component in your own online service? There is a thin line between adding values to your users and making them dissatisfied.

Social media usually put users who in various domains substantially exceed a regular user, in the spotlight. Through scrolling, we see fitness role models, fashion trend-setters, elite athletes and high-earners. The appear among the regular users. They are like everyone else, but better. Some studies documented that Instagram use results in increased compulsive buying of luxury products among young Asians. The cognitive mechanism that pushes them to take an action is upward social comparison.

Upward social comparison is a mechanism in which a person compares themselves to their better peers. This triggers motivation to put more effort and behave to catch-up with better performers, potentially letting individuals grow and progress. However, in some cases, it may be too difficult to catch up because the gap between you and the top performers is too large. What happens then?

In [1], we measured user satisfaction from their own performance when they were trading in a simulated stock market. One group received no information about the performance of others. The other group received information about performance about three very good investors. The second group was placed in a situation mimicking social trading apps. Social trading apps allow users to see other users' strategies and performance. In our experiment, traders who saw the top performer in the spotlight reported lower satisfaction from their own performance. Unfortunately, they did not perform better than the users who used the trading app without the social media component. However, they took more risky decisions and they were more active by making more trades with larger trading volumes. This might seem as wasted effort that does not lead to a desired effect.

When designing an interactive social component in their apps, companies should be mindful of the content and format of the user information that is available to other users. This should also be a concern for financial regulators such as FINRA Swiss Financial Market Supervisory Authority FINMA and European Commission.

Terms to remember: upward social comparison, user satisfaction, social trading app

References:

1. Andraszewicz, S., Kaszás, D., Zeisberger, S. & Hölscher (2023). The influence of upward social comparison on retail trading behaviour. Scientific Reports, 13, 22713, https://doi.org/10.1038/s41598-023-49648-3

Decarbonisation, securing global energy needs and undertaking sustainable actions stay at the forefront of global policy-making. Fostering behaviours aimed at preservation of common goods, such as clean air or clean water, can be challenging in a highly competitive and socialised world of today. Why is that? Because human brain deals with mixed reward signals between when facing sustainability goals and competition at the same time. 

In [1], we designed a gamified fishing experiment where participants earned Swiss Francs based on the number of fish they caught. The fish population was limited, with a fixed growth rate of 1.5, but it also declined due to external factors. Participants were randomly assigned to two groups: Group 1 was told that fish migrated to another lake (private context), while Group 2 was informed that two other people were also fishing (social context). In reality, the fish depletion rate was identical for both groups, with only the framing differing between a private or social scenario. In this game, with moderate resource consumption and outflow, the fish population would remain constant.

People in the social context depleted the resource faster by matching or exceeding the amount taken by others. In contrast, those in the private context took only enough to match the natural outflow, ensuring the resource lasted longer.

We conducted this study using functional MRI (fMRI) to measure brain activity while participants took resources. fMRI detects which brain areas use more oxygen, indicating where activity is highest. We found that the ventral striatum, a reward-processing brain region located underneath the cortex, responded differently in the social and private contexts. In the social context, the brain rewarded participants for taking more than their peers, while in the private context, it rewarded them for taking only as much as the natural depletion rate allowed.

Dealing with "the tragedy of the commons" is a social problem that requires behavioural interventions. The world leaders should not forget that many human behaviours are wired in the brain activity at the biological level. We explain this phenomenon to the young generations in [2].

Terms to remember: common goods, gamified experiment, private context, social context, Magnetic Resonance Imaging (MRI), functional MRI (fMRI), ventral striatum, cortex, the tragedy of the commons

References:

1. Martinez-Saito, M., Andraszewicz, S., Klucharev, V., and Rieskamp, J. “Mine or Ours? Neural Basis of the Exploitation of Common-Pool Resources.” Social Cognitive and Affective Neuroscience, vol. 19, no. 9, 2022, pp. 837-849, https://doi.org/10.1093/scan/nsac008.

2. Zappe, A., Martinez-Saito, M., and Andraszewicz, S. “What’s Mine? What’s Ours? How the Brain Thinks About Shared Resources.” New Discovery, Frontiers Young Minds - Neuroscience and Psychology, 2024.

Originally published on LinkedIn on 27.01.2025

Simulators have been frequently used for training purposes, but are less common in experimental research of dynamic decision-making. Along the technical development of computers, researchers started to develop microworlds - gamified computerised tasks that mimic real-life problems. At the ETH Zurich - Chair of Cognitive Science , we created the Zurich Trading Simulator (ZTS), which mimics a simple environment for trading financial assets. This is an example of a simulated experience. ZTS dynamically displays an asset price on a tick-by-tick basis, allowing for displaying news and manipulating pay-off structures (software description in [1]). ZTS is an app for oTree released under the Open Science licence. The code is available on GitHub (https://github.com/Zurich-Trading-Simulator/OtreeZTS) and the app is free to use by researchers and practitioners.

Using ZTS, we created a series of experiments in which we investigated the impact of social trading platforms on trading behaviour and trader's satisfaction. We found that upward social comparison, such as comparing oneself with traders with much higher earnings, results in higher trading activity and risk taking, while it reduces satisfaction from their own trading performance (detailed results in [2]). In another experiment measuring skin conductivity - a proxy for anticipatory arousal in goal-directed behaviour - when participants traded during a market bubble-and-crash scenario, we found that bodily signals can predict whether a person would earn or lose money on a stock market (detailed results in [3]). Overconfidence is conceived as one of the most prevalent biases in financial investing, but its effect on trading behaviour and performance is complex and indirect. We also found that incentivising people with bonuses changes their trading activity without changing their trading performance and salary outcomes, compared to other types of monetary incentives.

A curious reader interested in the psychological mechanisms that drive financial asset markets can find more information in this book chapter [4].

Terms to remember: microworld, simulated experience, upward social comparison, anticipatory arousal, goal-directed behaviour, overconfidence

References:

1. Andraszewicz, S., Friedman, J., Kaszás, D., & Hölscher, C. (2023). Zurich Trading Simulator (ZTS)—A dynamic trading experimental tool for oTree. Journal of Behavioral and Experimental Finance, 37, 100762. https://doi.org/10.1016/j.jbef.2022.100762

2. Andraszewicz, S., Kaszás, D., Zeisberger, S. & Hölscher (2023). The influence of upward social comparison on retail trading behaviour. Scientific Reports, 13, 22713, https://doi.org/10.1038/s41598-023-49648-3

3. Wichary, S., Allenbach, M., von Helversen, B., Kaszás, D., Sterna, R., Hölscher, C., & Andraszewicz, S. (2023). Skin conductance predicts earnings in a market bubble-and-crash scenario. https://doi.org/10.31219/osf.io/ybu8z

4. Andraszewicz, S. (2020). Stock Markets, Market Crashes, and Market Bubbles. In: Zaleskiewicz, T., Traczyk, J. (eds) Psychological Perspectives on Financial Decision Making. Springer, Cham. https://doi.org/10.1007/978-3-030-45500-2_10

Originally published on LinkedIn on 14.10.2024

I have heard professional traders from Zurich asking whether there is a scientific tool that that would help them deal with their biases and emotions when they make important investments. "Gut feeling" is a frequent driver of financial decisions. In a study described in [1]  simulating a stock market mimicking the Swiss stock exchange market, with over 100 students of a course about financial marek risks, we found that about 50% of the transactions were motivated by  investors' gut feeling. This gut feeling is linked to the "somatic marker hypothesis" - a theory which assumes that our body sends signals (markers) crucial in decision-making. Anterior Insula (AI) is a small brain structure situated inside of the brain, under the cortex. This structure plays a crucial role in sending such markers. Its activity can be measured using brain imaging techniques, but also with a very discrete and non-invasive method measuring skin conductivity. Skin conductivity increases when the skin's moisture increases together with sweating. Micro-sweating - a very small increase in sweating, non-observable with a bare eye - is related to the activity of anterior insula, signalling a person's increased arousal with increased skin conductivity. In cognitive science, arousal signals emotions and attention.

We conducted an experiment [3], in which we asked participants to trade on a simulated Zurich Trading Simulator [2] market during a bubble-and-crash scenario. We measured their skin conductivity level, while they were trading. We found that the skin conductivity level of the best performers predicted the change in the price trends, while the worst performers' attention decreased throughout the bubble-and-crash phase.  In this experiment, we used highly sensitive electrodes, wired to a measurement device. However, the quality of sensors in wearable devices constantly improves. In the future, it could be possible that a fitness tracker could help interpret bodily signals related to financial decisions.

Terms to remember: gut feeling, somatic marker hypothesis, anterior insula (AI), skin conductivity, micro-sweating, arousal, bubble-and-crash scenario, bodily signals

Publications:

1. Sornette, D., Andraszewicz, S., Wu, K., Murphy, R.O., Rindler, P. & Sanadgol, D. (2020). Overpricing persistence in experimental asset markets with intrinsic uncertainty. Economics: The Open-Access, Open-Assessment E-Journal, 14(20), 1-53. https://doi.org/10.5018/economics-ejournal.ja.2020-20

2 Andraszewicz, S., Friedman, J., Kaszás, D., & Hölscher, C. (2023). Zurich Trading Simulator (ZTS)—A dynamic trading experimental tool for oTree. Journal of Behavioral and Experimental Finance, 37, 100762. https://doi.org/10.1016/j.jbef.2022.100762

3. Wichary, S., Allenbach, M., von Helversen, B., Kaszás, D., Sterna, R., Hoelscher, C., & Andraszewicz, S. (2023). Skin conductance predicts earnings in a market bubble-and-crash scenario. https://doi.org/10.31219/osf.io/ybu8z