What Causes Forgetting? Decay Over Time or Interference with Other Learned Information?

For millennium, scholars and thinkers of all sorts — from scientists to men or women on the street — thought that memories simply faded with time.

Locke said:

"The memory of some men, it is true, is very tenacious, even to a miracle; but yet there seems to be a constant decay of all our ideas, even of those which are struck deepest, and in the minds of the most retentive; so that if they be not sometimes renewed by repeated exercise of the senses, or reflection on those kinds of objects which at first occasioned them, the print wears out, and at last there remains nothing to be seen."  John Locke quoted by William James in Principles of Psychology (p. 445, the 1952 Great Books edition, original 1891).

However, in the mid 1900's research by McGeoch (1932), Underwood (1957) and others found that memories can fade when what is learned interferes with other things learned. Previous things learned can interfere with current learning (proactive interference) and current learning can be interfered with by subsequent learning (retroactive interference).

The debate between decay and interference went on for over a century! Indeed, it paralleled the debate in physics over the property of light. Is it a wave or a particle?

The first ever photograph of light as both a particle and wave

In physics, the debate was so important that Albert Einstein won the Nobel Prize for the solution. Einstein's solution was simple. Light was BOTH a wave and a particle. The picture above is reported by Phys.org to be the first photograph demonstrating light's dual properties.

Now in the psychological research, we have the first experimental evidence that forgetting may be caused by BOTH decay and interference.

In a clever experiment, published just this month, Talya Sadeh, Jason Ozubko, Gordon Winocur, and Morris Moscovitch found evidence for both interference and decay.

Their research appears to be inspired, at least partially, by neuroscience findings. Here's what the authors say:

"Two approaches have guided current thinking regarding the functional distinction between hippocampal and extrahippocampal memories. The first approach maintains that the hippocampus supports a mnemonic process termed recollection, whereas extrahippocampal structures, especially the perirhinal cortex, support a process termed familiarity… Recollection is a mnemonic process that involves reinstatement of memory traces within the context in which they were formed. Familiarity is a mnemonic process that manifests itself in the feeling that a studied item has been experienced, but without reinstating the original context." (p. 2)

To be clear, this was NOT a neuroscience experiment. They did not measure brain activity in any way. They measured behavioral findings only.

In their experiment, they had people engage in a word-recognition task and then gave them either (1) another word-learning task, (2) a short music task, or (3) a long music task. The first group's word-learning task was designed to create the most interference. The longer music task was designed to create the most decay (because it took longer).

The results of the experiment were consistent with the researcher's hypotheses. They claimed to have found evidence for both decay and interference.


Every scientific experiment has caveats. Usually these are pointed out by the researchers themselves. Often, it takes an outside set of eyes to provide caveats.

Did the researchers prove, beyond the shadow of a doubt, that forgetting has two causes? Short answer: No! Did they produce some interesting findings? Maybe!

My big worry from a research-design perspective is that their manipulation distinguishing between recollection and familiarity is somewhat dubious, seemingly splitting hairs in the questions they ask the learners. My big worry from a practical learning-design perspective is that they are using words as learning materials. First, most important learning situations utilize more complicated materials. Second, words are associative by their very nature — thus more likely to react to interference than typical learning materials. Third, the final "test" of learning was a recognition-memory task that involved learners determining whether they remembered seeing the words before — again, not very relevant to practical learning situations.

Practical Ramifications for Learning Professionals

Since there are potential experimental-design issues, particularly from a practical standpoint, it would be an extremely dubious enterprise to draw practical ramifications. Let me be dubious then (because it's fun, not because it's wise). If the researchers are correct, that memories that are context-based are less likely to be subject to interference effects; we might want to follow the general recommendation — often made today by research-focused learning experts — to provide learners with realistic practice using stimuli that have contextual relevance. In short, teach "if situation–then action" rather than teaching isolated concepts. Of course, we didn't need this experiment to tell us that. There is a ton of relevant research to back this up. For example, see The Decisive Dozen research review.

Beyond the experimental results, the concepts of delay and interference are intriguing in and of themselves. We know people tend to slide down a forgetting curve. Perhaps from interference, perhaps from decay. Indeed, as the authors say, "it is important to note that interference and decay are inherently confounded."


The experiment:

Sadeh, T., Ozubko, J. D., Winocur, G., & Moscovitch. M. (2016) Forgetting patterns differentiate between two forms of memory representation. Psychological Science OnlineFirst, published on May 6, 2016 as doi:10.1177/0956797616638307.

The research review:

Sadeh, T., Ozubko, J. D., Winocur, G., & Moscovitch, M. (2014). How we forget may depend on how we remember. Trends in Cognitive Sciences, 18, 26–36.