When do you look something up, and when do you try to remember it?  One of the hallmarks of humans is our ability to store and access information in the environment around us as well as in our own central nervous systems. We have an external memory and an internal memory.

Many tasks involve moment-to-moment trade-offs between the two. The “soft constraints” theory posits that for a given task the extent to which your strategy draws more on external versus internal memory depends on the constraints of the task environment—in particular, how fast and easy it is to retrieve information from external memory. If your computer is really slow, you might opt to try and recall the title of that news article you read yesterday instead of checking your browser history.

Or suppose you are a teacher grading a stack of homework assignments.  You might start by comparing each student’s answers to a rubric you’ve printed out, looking back and forth between the student’s sheet of paper and the rubric sheet. But after grading the first few students’ assignments, you might begin to speed things up by memorizing the correct answers and comparing students’ answers to the contents of your memory rather than the rubric sheet. You’ve shifted from a strategy that relies more on external memory to one that relies more on internal memory.

A recent article by Patrick et al. in the Psychonomic Society’s journal Memory & Cognition reports evidence that your strategy selection is also influenced by prior experience.

A number of researchers have studied how people strategically manage the tradeoff between internal and external memory by running experiments using something called the Blocks World Task.  Here’s what happens in this task: you see a pattern of, say, 10 colored blocks in a 4×4 grid on the left side of the computer screen, and your job is to recreate that pattern in the empty 4×4 grid on the right side of the screen (the “workspace”).  You do this by clicking and dragging blocks to the workspace from a “resource window” at the bottom of the screen.  Once you’ve built the whole pattern of 10 blocks, you click on a “stop trial” button, and then you’ll start again with a new block pattern to build.

Sounds pretty straightforward, right?  But here’s the really clever part: researchers can introduce an “access cost” such that the target pattern you’re supposed to recreate (left side) is only revealed when the mouse cursor is over it. So you can’t look at the pattern and drag blocks from your resource window to the workspace at the same time. Furthermore, researchers can increase that access cost by introducing a delay between the moment you mouseover and the moment the pattern is revealed.  If you’ve ever waited a few seconds for a web page to load, you can imagine how annoying and tiresome this gets.  And research has shown that the bigger this delay gets (e.g., from 0 up to 3 seconds), the more participants try to memorize the pattern themselves (internal memory strategy) instead of frequently revisiting the pattern on the screen (external memory strategy).

Imagine again that you are the teacher grading homework, but you have to walk across the room every time you want to check the answer rubric. You’ll probably end up deciding to memorize the answers sooner than if the rubric is right in front of you.  The cost of accessing external memory influences the extent to which you adopt an internal strategy.  Building on this knowledge, Patrick et al. have now shown that peoples’ strategy adoption is also influenced by their previous training and experience.

In the experiment, participants completed 10 Block World Task trials, in which the target pattern appeared immediately upon mouseover.  The key manipulation was what participants did before those 10 trials.  There were three groups: one group did no previous trials, another group did 30 initial trials that worked just like the 10 critical trials, and the last group did 30 “high access cost” initial trials in which they had to wait 2.5 seconds for the target pattern to be revealed after mouseover.

That last group is the most interesting. For 30 trials they had to wait a few seconds every time they wanted to see the target pattern, so they came to rely more on their internal memory than participants in the no-delay practice group.  One way to measure this is the average number of blocks that they correctly placed after their first viewing of the target pattern: 3.45 versus 2.15.  High access cost trials led people to memorize more blocks on their first view than no-delay trials.

But here’s the key question: what did people do on the subsequent 10 critical trials, which were all no-delay?  Even though the 2.5 second delay was now gone, the people who had just done the high access cost trials still memorized more blocks upon their first viewing (2.86) than the people who had just done the no-delay trials (2.14).

These results make two critical points: First, for the high-access-cost group, their reliance on internal memory decreased (from 3.45 to 2.86) when the cost of using external memory decreased during the final 10 trials. Thus, the affordances of the task environment indeed influenced strategy.  Second, even so, those participants still relied more on internal memory than participants who hadn’t experienced the high-access-cost trials (2.86 vs. 2.14). Thus, there was also carryover of previously developed strategies. Therefore, both the environment itself and our prior experience influence our strategies for striking a balance between using our own minds and external aids.  This finding has implications for the soft constraints theory, which details the cost-benefit analysis made in selecting strategies to save both time and mental effort, but hasn’t yet included a role for prior experience.

Why does this kind of research matter?  The way we delegate tasks between our minds and the immediate working environment is important for productivity.  For example, some research has shown that internal-reliant strategies are more resilient to interruptions.  That’s not to say that internal strategies are always better.  What about possible costs to accuracy from greater reliance on internal memory?  In the study by Patrick et al., there weren’t any.  That is, there was no difference in the number of errors made by people in the high-access-cost group versus the no-delay group.  But I think it would be interesting to consider tasks where using external resources might yield higher accuracy, for example using calculators versus mental arithmetic under time pressure.

It’s also interesting to think of more naturalistic situations where there is an interplay between internal and external information stores.  For example, using a smartphone to look up information in the midst of a social interaction.  In addition to time costs (how fast is your phone’s connection?) and mental effort (how hard would you have to wrack your brain?), other factors may influence your strategy selection, such as your subjective confidence in your internal memory, or the importance of accuracy for the question at hand (“What was the name of that actress?” vs. “What date is that party?”).  I wonder if we might also see cultural and age cohort differences in things like peoples’ tolerance of delay in accessing external information, or the social acceptability of using a device during a conversation (cf. digital natives vs. digital immigrants).