Data and code for Malejka et al. (2021), "Correlation analysis to investigate unconscious mental processes". The present project focused on a particular domain of this literature, implicit learning. Studies conducted in this area try to determine whether we are able to detect regularities in our environment without awareness of those regularities. Finding evidence of awareness in these domains is important because it suggests that some degree of control may be available as well. In the present project we propose new methods for the study of unconscious learning. Many of the problems that we have detected in our previous research can be ameliorated by employing cutting-edge statistical analysis, including Bayesian and meta-analytic methods and model fitting. However, the validity of these approaches in the domain of implicit cognition remains untested.A consensus among researchers is that much of our behaviour is based on rather automatic processes we are barely aware of and over which we have little control. Research suggests that exposure to subtle cues can have dramatic effects on our decisions. For instance, asking people to provide the last 2 digits of their social security number biases how much they are willing to pay for products and commodities. Similarly, according to some researchers, people are more likely to be impolite and disrespectful if they have been exposed to words related to rudeness while solving anagrams. Another line of research suggests that we take many of our (important) decisions when distracted and thinking about other things and that this 'unconscious thought' process actually improves the quality of our decisions. These studies pertain to a larger area of research usually called 'implicit cognition', which explores how unconscious mechanisms contribute to cognitive processes including perception, learning, memory, and decision making. This area of research has attracted a great deal of attention from the media and features frequently in popular science books, blogs, and documentaries. Some authors have even suggested that parts of this research could be used to improve our decisions in different domains at a societal level (for example, in health behaviour and pension planning). The present project focuses on a particular domain of this literature, implicit learning. Studies conducted in this area try to determine whether we are able to detect regularities in our environment without awareness of those regularities. In other words, these studies address whether we can learn something without realising that we are indeed learning it. In recent years there have been thousands of demonstrations of implicit learning effects in the scientific literature and, not surprisingly, this literature has become increasingly influential in all areas of psychology, with an important impact in our understanding of human cognition and psychopathology. Unfortunately, our previous research suggests that much of this evidence is undermined by fundamental methodological problems that preclude any strong conclusions about the reliability of unconscious learning effects. We have shown that many of these studies find unconscious learning because researchers use weaker methods to assess whether people are conscious of what they have learned than to assess whether learning has taken place. Naturally, this implies that learning is easily detected but awareness is not, which creates the illusion that learning has taken place unconsciously. Finding evidence of awareness in these domains is important because it suggests that some degree of control may be available as well. In the present project we propose new methods for the study of unconscious learning. Many of the problems that we have detected in our previous research can be ameliorated by employing cutting-edge statistical analysis, including Bayesian and meta-analytic methods and model fitting. However, the validity of these approaches in the domain of implicit cognition remains untested. A second goal is to conduct a large-scale exploration of the prevalence and magnitude of these problems. Our previous studies have focused on a very particular effect studied in implicit learning research ('contextual cueing'). We suspect that many of these problems transcend this domain and affect a large proportion of current studies on implicit learning. The potential impact of this assessment is difficult to overestimate. Finally, we will set up a collaboration with other international laboratories working on this topic to gather the largest and most sensitive data set of implicit learning effects available so far. This data set will be publicly available for all researchers, which will make it a fundamental resource for the study of unconscious cognitive processes for many years to come.

Computer simulation