In order to experience the world as we do, we have to keep track of a great deal of information.  Information about where an object was or the last thing someone said to us lingers in our minds even after the object or statement is no longer present.  This is particularly true in the case of streams of information, such as melody or speech.  When we hear a melody, we are able to somehow hold onto the notes that preceded the current one in order to experience the ensemble as music.  This is a form of memory.  It is short-lived (as you can see when you try to listen to very, very slow music), and we are often not even aware that we are remembering something.  This kind of memory simply changes the way we respond to future events.

This complex phenomenon can be simplified for the purposes of research.  In our lab, we use streams of visual (rather than auditory) information.  Instead of notes and words, we use red and blue circles.  But people show memory for these objects as well.  When they see a type of circle they have just seen (the same color for instance), they process it more quickly.  This means that when they respond to the second circle, they will have faster reaction times.  So the memory system here serves to change the speed of processing of future objects based on the ones people have already seen, which we see in experiments as faster and slower reaction times.

There is a slightly more complex version of this effect.  We know that people are quick to respond to circles they have seen before, and that they are slower when they haven't just seen the same kind of circle.  But what if the circles differ in more than one way?  Rather than changing color alone, we can also change circles' positions on the experiment screen.  Instead of only two kinds of objects (red or blue circles), then, we would have four types of objects: red-high, red-low, blue-high, and blue-low.  This would also mean that each new circle could relate to the previous circle in one of four ways.  It could be the same color and the same position (all same), it could be a different color and a different position (all different), or it could be the same color and a different position, or a different color and the same position.  People reliably respond faster to certain transition types: specifically, same-color/same-position and different-color/different-position (that is, to an object exactly alike or exactly opposite of the one before).  When only color or position changes, the second object is processed more slowly.  Plotting the time it takes for people respond to all four transition types gives us the figure below. (Note that a z-score is simply a way of standardizing the reaction times.  Lower numbers indicate faster responses.)

This X (or "crossed interaction") is the marker of a fully functioning implicit memory system.  It means that some part of the nervous system is changing its behavior based upon previously encountered stimuli.  It is doing this below the level of conscious awareness -- people in our experiments have only the barest idea that they might be responding differently to a particular sequence of objects than to another.

In this series of studies, we are interested in discovering whether it is possible to change the strength of this memory system.  We already know that it diminishes with time.  By the time 2-3 seconds has passed, the cross will have mostly disappeared, indicating that the memory system is no longer functioning (at least, not at full strength).  For more information on temporal decay, see our article on Implicit Working Memory.

However, this does not tell us if there is anything we can do on a short-term scale to change implicit memory strength.  One way to do this might be to introduce grouping cues.  If you know that two items belong together, it makes sense to compare them and consider how they relate to one another.  But if you think that two objects are in different groups, there is much less need to do so because you can assume that they are probably unrelated.  This makes sense ecologically, but it is necessary to run experiments to discover whether human memory actually works this way. 

We tested this idea using rhythm.  This is a two-note sequence followed by a long rest, repeated five times: Listen

The two-note pair will naturally appear as a group.  The long rest serves as a boundary separating this group from past and future groups.  This is an auditory rhythm, but we can also create rhythm with our visual system.

If we present our high and low, red and blue circles at this rhythmic pace (circle-circle-rest-circle-circle-rest...), we might expect people to group them in the same way.  We would know if this sort of grouping were occurring in memory by looking at the strength of the implicit memory cross.  Since the memory cross depends on which type of circle the viewer last saw (before the one they actually responded to), we can divide up our experiment depending on whether the last circle was in the same group, or if there was a rest between it and the current circle.  Implicit memory should be strong within groups, so we would expect to see a cross there.  However, memory should be weak for items from different groups, so we would expect the cross to be diminished or gone altogether at the boundaries (rests).

As expected, the implicit memory cross completely collapsed at our event boundary.  This indicates that the memory system function is momentarily destroyed just at this point.  Within groups, however, implicit memory is functioning normally, as seen by the X.

This experiment demonstrates that it is possible to modulate implicit memory using nothing but event structure.  We are currently conducting additional studies to further explore and understand the full extent of this phenomenon.