We’re all familiar with the experience of going to another room and forgetting why we’ve done so. The problem has been largely attributed to a failure of attention, but recent research suggests something rather more specific is going on.
In a previous study, a virtual environment was used to explore what happens when people move through several rooms. The virtual environment was displayed on a very large (66 inch) screen to provide a more immersive experience. Each ‘room’ had one or two tables. Participants ‘carried’ an object, which they would deposit on a table, before picking up a different object. At various points, they were asked if the object was, say, a red cube (memory probe). The objects were not visible at the time of questioning. It was found that people were slower and less accurate if they had just moved to a new room.
To assess whether this effect depends on a high degree of immersion, a recent follow-up to this study replicated the study using standard 17” monitors rather than the giant screens. The experiment involved 55 students and once again demonstrated a significant effect of shifting rooms. Specifically, when the probe was positive, the error rate was 19% in the shift condition compared to 12% on trials when the participant ‘traveled’ the same distance but didn’t change rooms. When the probe was negative, the error rate was 22% in the shift condition vs 7% for the non-shift condition. Reaction time was less affected — there was no difference when the probes were positive, but a marginally significant difference on negative-probe trials.
The second experiment went to the other extreme. Rather than reducing the immersive experience, researchers increased it — to a real-world environment. Unlike the virtual environments, distances couldn’t be kept constant across conditions. Three large rooms were used, and no-shift trials involved different tables at opposite ends of the room. Six objects, rather than just one, were moved on each trial. Sixty students participated.
Once again, more errors occurred when a room-shift was involved. On positive-probe trials, the error rate was 28% in the shift condition vs 23% in the non-shift. On negative-probe trials, the error rate was 21% and 18%, respectively. The difference in reaction times wasn’t significant.
The third experiment, involving 48 students, tested the idea that forgetting might be due to the difference in context at retrieval compared to encoding. To do this, the researchers went back to using the more immersive virtual environment (the 66” screen), and included a third condition. In this, either the participant returned to the original room to be tested (return) or continued on to a new room to be tested (double-shift) — the idea being to hold the number of spatial shifts the same.
There was no evidence that returning to the original room produced the sort of advantage expected if context-matching was the important variable. Memory was best in the no-shift condition, next best in the shift and return conditions (no difference between them), and worst in the double shift condition. In other words, it was the number of new rooms entered that appears to be important.
This is in keeping with the idea that we break the action stream into separate events using event boundaries. Passing through a doorway is one type of event boundary. A more obvious type is the completion of an action sequence (e.g., mixing a cake — the boundary is the action of putting it in the oven; speaking on the phone — the boundary is the action of ending the call). Information being processed during an event is more available, foregrounded in your attention. Interference occurs when two or more events are activated, increasing errors and sometimes slowing retrieval.
All of this has greater ramifications than simply helping to explain why we so often go to another room and forget why we’re there. The broader point is that everything that happens to us is broken up and filed, and we should look for the boundaries to these events and be aware of the consequences of them for our memory. Moreover, these contextual factors are important elements of our filing system, and we can use that knowledge to construct more effective tags.