I am finally back in my office after a short holiday and after doing some teaching. I am glad to announce that I submitted my first year report and successfully defended it. That is the reason why, I haven’t posted anything in a while, but as soon as I do more work again, that can be uploaded, I will do that.
The results reported in my first year report are very promising. They show that there is a reliable U-shaped relationship between object/location expectancy and memory performance. Further analysis (link) among other things showed that this is also the case when only items are analysed that are expected in a kitchen so that there is no confounding influence from the general expectancy of the objects.
To show that this effect is not depending on a specific combination of objects and locations, I will run a third experiment (schemaVR3), where multiple sets of object/location combinations are used instead of just one for each experiment as it was done in schemaVR1 and schemaVR2. This is important because at the moment it is still conceivable that the results arise from unfortunate sampling of objects and locations. An alternative to using a number of sets, would be to present all objects at random locations for each participant. The problem with that approach is that one could not analyse the correlation between the mean object/location expectancy and the number of familiar and remember responses because averaging result in an expectancy around zero. This however is important because we found in schemaVR2 that unexpectedness is associated with more remember responses and expectedness is associated with more familiar responses. Using a small number of sets will furthermore increase our power to detect a meaningful correlation because it will increase the number of data points from twenty to a multiple of that number.
Also because the 3AFC and the 3D location recall task showed the same pattern, we decided to drop the 3AFC as we will gain more freedom in selecting suitable sets. We also considered to replace the eight non-kitchen items, which greatly vary in terms of their general expectancy, with eight kitchen items, but as the additional analysis (link) is showed that is not necessary. I will also try to write a function that tracks movement of the participants and their head orientation for future analysis. In a next step, I will look in the possibility to detect what objects are visible at any given time point.
Similar to the object/location selection for the previous experiments (schemaVR1 and schemaVR2), I will find the combination with the best spread and then check whether this combination works in the virtual environment and then do the same with the second best until I get five viable sets. As we are dropping the 3AFC, there is no need to find suitable foils.
In addition to preparing the third experiment, I will prepare the data and all scripts as well as the manuscript for publication. As soon as, I am ready I will upload the scripts and the experiments to this GitHub repository.