A new perspective on human judgement and decision making as optimal: A framework for behaviour change

Lead Research Organisation: University of Manchester
Department Name: Psychological Sciences

Abstract

Psychology and Economics typically characterise human judgement & decision-making (JDM) as flawed. JDM is commonly described as irrational, biased or sub-optimal and there are many observations that appear to support this viewpoint. For example, when asked to make a judgment about the likely outcome of a coin toss after a long sequence of previous tosses coming up heads, many people expect a tail on the next toss, when in reality previous tosses do not change the 50:50 head:tail chance. This misperception is known as the Gamblers Fallacy (GF). If, however, we accept that humans have a narrow focus of attention and a limited memory capacity for say, only the last 5 tosses, this will limit what can be learnt from the long sequence. This limited experiential window means that, mathematically, not all sequences of a given length have the same chance of occurring in a longer (but finite) sequence and as a result the gamblers' expectation of a tail after a run of heads can be shown to be entirely reasonable.
In this project we aim to show that if we take into account the limits on human cognitive capacity or the limits imposed by the environment, human JDM is not flawed and is in fact optimal given those constraints. We will demonstrate this for two examples, the GF above and 'Preference Reversals' (PR) another class of phenomena in which JDM is commonly described as biased. Further, we will show that if we change cognitive or environmental constraints, humans will adapt and remain optimal given the new constraints. This will allow us to test the consequences of manipulating the causal determinants of human JDM performance and therefore will allow for better policy development. To do so we will conduct behavioural experiments as well as formulate and evaluate computational models of our experimental tasks.
The project comprises two themes running mostly in parallel over the 18 months project. Theme 1 will investigate human Perception of Randomness and the associated GF phenomenon. The ability of humans to perceive structure in the environment and differentiate it from its counterpart, randomness, is at the heart of our ability to make causal links between events. For this reason, studying perception of randomness is vital for our understanding of how humans understand the world. Using the GF and our theoretical analysis, we will run a series of behavioural experiments to explain GF as optimal under constraints - namely the short-term memory limitation as a proxy for an individual's window of experience. We will use computational models to test if the observed behaviour is optimal under such constraints. Further experiments will simulate interventions that modify the key constraints, for instance, temporarily reducing the size of the window of experience by varying cognitive load. We hypothesise that behaviour will change in line with model predictions, and remain optimal. This would be a significant finding and a strong test of the proposed framework.
Theme 2 will investigate context effects and Preference Reversal (PR). PR occurs when the addition of further choices changes a person's preference, even when the new choices are worse. Although PR is used as evidence of human irrationality, our pilot work shows that introducing new choices provides additional information about priors, such that PR becomes an optimal behaviour. We will manipulate the distribution of choice problems and measure adaptation to the different task distributions by measuring changes in the reliance on context.

Planned Impact

The project will develop a causal model of human judgement & decision-making (JDM) that goes beyond describing behaviours to predicting behaviours by understanding why people make decisions, identifying key environmental or human constraints on JDM and evaluating their impact. The project aims to demonstrate that if the constraints imposed by the environment and the limitations of the human cognitive system are taken into account, human judgment and decision making (JDM) are optimal. This is a major shift from the current dominant perspective in Psychology and Behavioural Economics which suggests JDM is irrational, biased or sub-optimal. In doing so the project will replace the descriptive accounts of JDM by developing a novel, causal and predictive framework that facilitates the design and testing of interventions to bring about behavioural change. The model will allow the testing of behavioural change scenarios and prediction of the outcomes. The project also aims to act as a stimulus for novel avenues of investigation across a range of fields in Psychology, Economics and Social Science, with particular emphasis on the constraints impacting on real-life decision making. Practically, it would inform a principled approach to policy change and strategies for behavioural change interventions in a range of contexts, for example to remove the cognitive barriers against sustainable consumption and healthy life styles or to promote better personal finance/financial investment decisions.

We suggest that previous approaches to induce behaviour change can be less complete than that offered by our account. For example, if a behaviour such as a preference reversal is not understood (and we contest that it is not) then although it is possible to modify behaviour by adding dominated choices to change the context, this does not explain why the behaviour change occurs. Furthermore, people may rapidly adapt to the changed priors that are a consequence of changing the relationship between context and outcome. If, instead, the causal determinants of the behavior are understood then there is considerably greater scope for designing interventions by modifying key constraints and assessing the impact of those interventions.

Specifically, the impact can be split into the following two categories:
1. Academic: The project aims to transform the perspective in the Psychology & Behavioural Economics literature, by placing the emphasis on optimal (under constraints) rather than suboptimal behaviour. It will provide the basis for a stimulating a new agenda for social science research and offer a unifying framework that brings together a disparate body of phenomena in JDM research. To realise this impact we will disseminate our work at national and international conferences, propose symposia at these conferences and write further funding applications.
2. Socio-economic: The project will provide a rigorous framework for designing and testing behavioural change interventions. This will lead to greater understanding the causal factors underpinning a range of decisions with socio-economic impact, for example sustainable consumption and healthy life styles or to promote better personal finance/financial investment decisions. To realise this potential we will establish contact with the Behavioural Insights Team at the cabinet office ('the nudge unit') and the financial institution AXA to disseminate our results and promote the proposed framework.

Publications

10 25 50

publication icon
Farmer G.D. (2015) The attraction effect in motor planning decisions in Judgment and Decision Making

publication icon
Farmer GD (2017) Who "believes" in the Gambler's Fallacy and why? in Journal of experimental psychology. General

publication icon
Farmer GD (2017) The Effect of Expected Value on Attraction Effect Preference Reversals. in Journal of behavioral decision making

publication icon
Farmer, G (2015) The attraction effect in motor planning decisions in Judgement and Decision Making

publication icon
Howes A (2016) Why contextual preference reversals maximize expected value. in Psychological review

publication icon
Warren PA (2018) A re-examination of "bias" in human randomness perception. in Journal of experimental psychology. Human perception and performance

 
Description The dominant perspective in psychology and behavioural economics suggests that human judgement and decision making (JDM) is fundamentally flawed (biased/irrational/sub-optimal). This notion has also been popularised through the media (http://www.bbc.co.uk/news/science-environment-26258662) and popular science books (e.g. Thinking fast and Slow by Daniel Kahneman; Predictably Irrational by Dan Ariely, etc). Our transformative award has allowed us to build a compelling argument that this perspective is incorrect. Contrary to the idea of flawed JDM, we provide evidence that humans actually perform as well as possible, provided certain cognitive limitations are taken into account. Under this framework we have developed predictive accounts (and supporting data) for two examples of so-called cognitive bias - phenomena which have presented strong challenges to the notion of rationality in human JDM: i) Contextual Preference Reversals (CPR) (preferences for one of two options are changed by addition of seemingly irrelevant third option); ii) The Gambler's Fallacy (GF) (tendency to believe a head becomes more likely after a long run of tails).

Primary achievements:
1. We developed a model of human decision making incorporating human unreliability in option evaluation that makes the best decisions it can given this constraint. This model predicts all the contextual preference reversal phenomena and moreover suggests seemingly sub-optimal preference reversals are actually required in order to make the best decision. To our knowledge this is the first genuinely causal model of the CPRs. As testament to the potential of this work to transform the field, the paper describing this model is in press at Psychological Review.

2. We conducted CPR experiments with human participants (in multiple decision making contexts) which are consistent with our account of (cognitively bounded) optimal human decision making in the face of unreliable evaluation of options.

3. We showed that behaviour in randomness perception experiments is moderated by experience of recently observed environmental statistics, in a manner consistent with our previous account of human randomness perception (Hahn & Warren, 2009). In this account we suggested limitations in short-term memory and/or attention constrain how we experience random events and this would give rise to the GF (Hahn & Warren, 2009). We also suggest the literature has been too hasty to classify participants as prone to GF; the majority of our participants do not express explicit GF-consistent beliefs, even though they exhibit a tendency to generate responses that are implicitly consistent with GF. We suggest such data do not provide robust evidence of erroneous beliefs, but rather sensitivity to environmental statistics

In summary our work provides theoretical and empirical insight to motivate a broad new research agenda in psychology and behavioural economics. We suggest it will be possible to reclassify other irrationality phenomena as rational under appropriate constraints in a unified predictive framework. Through a process of constraint identification and measurement together with modelling, our approach should enable accurate prediction of individual and collective behaviour. Ultimately this should lead to better models of JDM which can be used to guide policy design across a range of contexts.
Exploitation Route Non-Academics:
Our models could be developed to predict outcomes of policy changes and/or design interventions to promote advantageous behaviour across a broad set of domains (sustainability, retail/marketing, health, etc). Our recent workshop on randomness perception enabled us to present our findings to stakeholders. These included Prof. Nick Chater, an advisory board member for the UK Behavioural Insights Team (http://www.behaviouralinsights.co.uk) and we also had extremely positive feedback from a representative of /KL.7, Denmark's leading behavioural design agency (http://kl7.dk/en), which provides behavioural insight consultancy for both private and public bodies (e.g. Danish Road Directorate, Carlsberg). We envisage that such groups are interested in our approach in the long term to help in the design of interventions to change behaviour.
Academics:
We hope our work should lead to a revision of the accepted view that human JDM is flawed. By demonstrating that two phenomena providing compelling evidence for irrationality might actually be considered rational under appropriate constraints, we suggest that all such phenomena require re-evaluation. We therefore hope the field will therefore adopt a more principled approach to assessing rationality. Our findings also challenge cognitive neuroscientists to identify neural correlates of constraints on performance (e.g. Clark et al., PNAS, 2014)
Sectors Education,Environment,Financial Services, and Management Consultancy,Healthcare,Government, Democracy and Justice,Retail

 
Description The output of this 18 month-long project was primarily theoretical and it has taken longer than anticipated for research findings to reach the public domain. We have now provided compelling evidence in support of our primary hypotheses and developed causal models of human decision making behaviour. As testament to the strength of this work and potential for transformative impact, note that it is currently in press at Psychological Review. However, it is still too early for these findings to translate to direct non-academic economic or societal impacst. In addition due to the lack of a dedicated follow-on funding scheme after the end of our award the progress we made was halted. We will now look to engage non-academic stakeholders and make them aware of our research findings. In particular Dr. Warren is in contact with Simon Bentholm a partner at kl.7 a Danish Behavioural Design company (http://kl7.dk/en). Mr. Bentholm attended the randomness workshop we hosted at Birkbeck University and has subsequently been in contact with Dr. Warren to ask him to present our findings to the kl.7 team.
First Year Of Impact 2015
 
Title Computationally rational model of human decision making 
Description This is a model of human decision making under cognitive constraints which uses the Bayesian framework. Briefly, noisy estimates of expected value and combined with supplementary noisy ordinal attribute information in a Bayes optimal fashion. The model is able to predict all the standard contextual preference reversal effects (Atraction, Compromise, Similarity, Phantom Decoy effects) and is consistent with several extant data sets in the literature. The model has been developed throughout the course of the grant. 
Type Of Material Computer model/algorithm 
Provided To Others? No  
Impact The model forms the basis of an extensive paper which is currently under review at Psychological Science 
 
Description Invited presentation to Danish Behavioural Insights company KL7 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Industry/Business
Results and Impact I gave an invited presentation to the Danish Behavioural insights company KL.7 (http://kl7.dk/en/). Simon Bentholm was an attendee at the randomness workshop organised at the end of our grant and we have kept in contact since. When the two primary publications from this grant appeared (the Howes at al. (2016) Psych Review paper and Farmer et al. (2017) JEP: General paper) Simon asked me to present our findings to his team. 10-15 people were present and we have stayed in contact since the presentation. I am hoping that in future this may lead to a collaboration and plan to discuss the possibility of adding KL.7 as partners on a CASE studentship application.
Year(s) Of Engagement Activity 2016
 
Description Poster presentation at Cognitive Sciences Society Meeting, Quebec, Canada 2014 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Other academic audiences (collaborators, peers etc.)
Results and Impact Considerable interest in the results presented on the poster. Strong feedback.

Numerous email requests for copies of poster and associated publications,
Year(s) Of Engagement Activity 2014
URL http://cognitivesciencesociety.org/conference2014/index.html
 
Description Talk at Decision Making Bristol, September 2014 by Andrew Howes: "Computationally rational decision making" 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Other academic audiences (collaborators, peers etc.)
Results and Impact Talk presented by A. Howes at Decision Making Bristol, September 2014.
The cross-disciplinary Decision-making research group at the University of Bristol hosted a four day conference to bring together theoreticians and experimentalists on 9-12 September 2014. This conference aimed to embrace a breadth of research, appealing to those interested in all aspects of decision making. This conference was attended by some of the most influential researchers studying aspects of human decision making (Gerd Gigerenzer, Tom Griffiths, Nick Chater, Daniel Wolpert).



Talk sparked discussion on the notion of computational rationality
Year(s) Of Engagement Activity 2014
URL http://www.bris.ac.uk/decisions-research/conference2014/
 
Description Talk at Decision Making Bristol, September 2014 by George Farmer: The attraction effect is rational given uncertainty in expected value calculation 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Other academic audiences (collaborators, peers etc.)
Results and Impact Talk presented by G. Farmer at Decision Making Bristol, September 2014.
The cross-disciplinary Decision-making research group at the University of Bristol hosted a four day conference to bring together theoreticians and experimentalists on 9-12 September 2014. This conference aimed to embrace a breadth of research, appealing to those interested in all aspects of decision making. This conference was attended by some of the most influential researchers studying aspects of human decision making (Gerd Gigerenzer, Tom Griffiths, Nick Chater, Daniel Wolpert).

The talk sparked considerable interest from the audience on our novel results on decision making.

Dr. Farmer received many follow up emails from researchers interested in our results and will inform these individuals when the results are published.
Year(s) Of Engagement Activity 2014
URL http://www.bris.ac.uk/decisions-research/conference2014/
 
Description Talk at Randomness and Structure Workshop (London, Feb 2015) by Paul Warren: "Assessing the bias in human randomness perception" 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Other academic audiences (collaborators, peers etc.)
Results and Impact Talked sparked questions and discussion afterwards. Particular interest from non-academic stakeholders interested in behaviour change.

The talk attracted extremely positive (verbal and email) feedback from a representative of /KL.7, Denmark's leading behavioural design agency (http://kl7.dk/en), which provides behavioural insight consultancy for both private and public bodies (e.g. Danish Road Directorate, Carlsberg). We envisage that such groups are interested in our approach in the long term to help in the design of interventions to change behaviour
Year(s) Of Engagement Activity 2015