by Gerald Epling

Is it possible for memory to increase over time after you have seen something or someone?  The short answer is, “yes”.  The long answer teaches us something about how we are put together and how the mind works.

In order to get at the activity of the mind when we see and learn faces, I decided to engage basic, primitive processes of the mind.  Because primitive processes spontaneously occur, there would be no way to exclude one process while engaging the other.  These processes would have to be engaged preferentially with different learning tasks.  Two primitive processes were selected for the empirical study of face recognition.  The first primitive process gives us the ability to see and learn something that is not immediately related to other items.  This process is called item-specific processing.  The second process is relational processing.  Clearly, item-specific processing occurs before relational processing.  Because, the activity of relating different items requires two or more items.

Seeing faces is a common human experience.  Over a lifetime people tend to develop expertise in sorting faces into categories.  We develop an ability to quickly recognize a face as belonging to specific group, such as women, men, quiet, outgoing, and so forth.  It is reasonable to expect that people who volunteered for the study would automatically engage relational processing, along with a little item-specific processing.  The amount of item-specific processing could be just enough be able to sort faces into a groups based upon common traits or characteristics.

Encouraging relational processing was accomplished by asking the participant to set up a number of different categories of people in their mind, and then to sort the faces into these different categories based upon perceived similarity or compatibility.

Item-specific processing was promoted by asking participants to sort faces by distinctiveness.  Most distinctive would be a 5, and less distinctive would be a lower.

Time between sorting faces and taking a test of the ability to recognize the faces was varied.  This was because I suspected that there is an optimum time between seeing a face and taking the face recognition tests. The time between study and tests allowed time for sensory and perceptual experiences to be consolidated into memory.

The results are shown graphically in the last section of this article.  The chart in figure 1 shows the responses of the two different groups when there was no delay between studying the faces, test 1, and test 2.  The chart in figure 2 shows the accuracy of face recognition when two minutes separated the study of the faces, test 1, and test 2.  The chart in figure 2 shows recognition accuracy with a 4 minute delay between studying the faces, test 1, and test 2.



As it turned out, relational processing produced good recognition accuracy in all conditions by the end of test 2.  There was an increase in accurate recognition for the relational processing group with no delay between study, test 1 and test 2 (p=.02).

The item-specific processing group, our distinctiveness sorters, showed an increase in accurate recognition between test 1 and test 2 when the delay time between study and tests was 2 minutes (p = .007).

With a 4 minute delay there was no significant difference between test 1 and test 2 for either group.

Non-significant values for interaction with no delay, 2 minute delay, and 4 minute delays were p = .18, F=1.95; p =.067, F= 3.94 and p=.39, F=.77 respectively.

There was no significant difference in the time that the groups took to study or recognize the faces.  Without a difference in time, and a difference in levels of recognition over time, it is clear that the two groups were doing things differently.


What does this mean?

Finding differences in the faces likely led people in the item-specific processing group to look more deeply into each face.  With just one category, participants had little relational processing to do.  Only the level of distinctiveness needed to be evaluated.  This allowed the item-specific processing group to spend more time on item-specific processing than the other group.

In contrast with item-specific processing, sorting faces into compatible groups likely required more relational processing and less focus in individual differences than the item-specific task.

Our ability to recognize people and things in our lives is made possible with three primitive processes, item-specific processing, relational processing, and a process that stores things in memory.  I call this third process, essential processing.  Essential processing is how we store the memories that we want to have or memories that are triggered by events occurring at levels below our conscious awareness.

Item-specific processing is showcased in the results that occurred with 2 minutes between study and test.  There was increase in memory between test 1 and test 2, when study and tests were separated by a 2 minute distractor task.   This suggests that the brain and mind take about 2 minutes for distinctive elements of a face to be consolidated and stored in memory.


How to use this information

If you want people to deeply learn new faces then it is a good idea to encourage item-specific processing of faces.  Item-specific processing leads to knowing.  One way to promote item-specific processing is to make pictures of faces easily available.  We see this with baseball cards and football cards.  The method also works with playing cards, as our Armed Forces found in Iraq over a decade ago.

If you only need to know enough to sort a face, or any other item, then relational processing is good enough.  The memory will not run as deep.  This allows you more time and mental room to do other things with your life.



Six different groups of people participated in the study.  Each line in the figures represents the experiences of one of the six different groups of people who volunteered.

In figures 1, 2, and 3 accurate face recognition is measured along the vertical axis.  The results of test 1 and test 2 are plotted along the horizontal axis.

Figure 1 shows an increase in memory over time for faces that were viewed with an eye toward organizing the faces into groups of compatible people.

Figure 1. No delay between study, test 1, and test 2.

Figure 1. No delay between study, test 1, and test 2.


Figure 2 shows an increase in memory over time for faces that were viewed with an eye toward the distinctiveness of individual faces.

Figure 2. Two minute delays between study, test 1, and test 2.

Figure 2. Two minute delays between study, test 1, and test 2.


Figure 3 shows no significant results.  The four minute delay was long enough to allow little change in memory over time.


Figure 3. Four minute delays between study, test 1, and test 2.

Figure 3. Four minute delays between study, test 1, and test 2.