Today, I finished the data collection for schemaVR1. I collected 17 instead of 16 participants because I wanted to show the experiment to someone who then agreed to formally take part in this experiment.
The two major the results of this experiment are that the locations of objects that are generally unexpected in a kitchen are remembered better (measured by 3D location recall and 3AFC location recognition) than the locations of objects that are expected in a kitchen. This can already be seen in the interim analysis (N = 9) that I uploaded here. The second result, which was only present as a ‘trend’ in the interim analysis, is that there is a positive quadratic relationship between memory performance and object/location expectancy once I control for the general expectancy of the objects .
I will, probably by the end of the next week, upload a ‘final’ analysis of the data that also will be part of my first year report. Before I can do that though I want to incorporate the suggestions by my supervisor. On my to-do list for that analysis are the following points:
- Analysing just the twelve expected objects to investigate whether this quadratic effect is just driven by the unexpected objects. It could also be the case that the microwave that is in the mid-range in terms of its object/location expectancy is driving the effect because memory performance for it is quite poor.
- Because of this, I will also run the analysis without the microwave.
- In the interim analysis, general expectancy is entered as a continuous factor, but I originally planned the study as a factorial design. That and the fact that objects nicely cluster in terms of the memory performance in addition to their expectancy is the reason why I will model general expectancy as a categorical factor (expected vs. unexpected).
- I will create plots showing the residuals for the models that include covariates to show the statistical effects.
- I will also plot the individual slopes of each participant to see how consistent the effect is across participants.
- In addition to the recall and 3AFC, I also looked at time need to place an object because I saw that in a paper that I read recently and found some interesting relationships. To see what this variable adds, I will correlate it with memory accuracy.
- To check whether the quadratic relationship follows an indeed a U-shape, I will make a median split and compare the slopes.
The results described above motivated the following changes for the second experiment (schemaVR2). I selected a new combination of objects and locations. The script for that can be found here. In schemaVR2, the spread of generally expected objects in term of their object/location expectancy will cover the whole range. The SLIMM model (see van Kesteren et al., 2012) postulate that both expected and unexpected locations are associated with better memory. However, two processes are according to that model responsible for that with different involvement of the mPFC and the HPC. To show that two qualitatively different processes underlie this, I now ask the participants for recollection and familiarity judgements after each trial. My prediction is that unexpected objects are associated with recollection, while expected objects are associated with familiarity. Furthermore, I again predict to see the U-shaped relationship between object/location expectancy and memory performance.
van Kesteren, M. T. R., Ruiter, D. J., Fernández, G., & Henson, R. N. (2012). How schema and novelty augment memory formation. Trends in Neurosciences, 35(4), 211–219. https://doi.org/10.1016/j.tins.2012.02.001