The intuitive model of forgetting is decay: memories fade like ink in sunlight, and time is the cause. It is a comfortable picture and it explains less than it seems to.
If time alone caused forgetting, then two people with the same gap since learning should forget at the same rate. They do not. What fills the gap matters enormously, and one of the more striking findings in memory research is that sleeping after learning preserves more than an equivalent period awake, because less happens in between.
Interference theory says the same thing more generally: memories compete for retrieval, and forgetting is mostly the loss of that competition.
The two directions
| Type | What interferes | Everyday version |
|---|---|---|
| Proactive interference | Old learning blocks new | Still typing last year's password |
| Retroactive interference | New learning blocks old | Learning Spanish damages your Italian |
Proactive interference is why the second thing is harder than the first. Your old parking space, your old phone layout, the old version of a process at work: the established memory keeps winning because it is stronger and has been retrieved more often.
Retroactive interference runs the other way and is the reason cramming several similar subjects into one evening works so badly. The third topic degrades the first, and the more alike they are, the worse the damage.
Similarity is the aggravating factor
Interference scales with how similar the competing memories are. Two entirely different subjects barely touch each other. Two subjects that share vocabulary, structure, and format will tear each other apart.
This has an immediate consequence for anyone studying: the material most at risk is the material you most need to distinguish. Two drug names that differ by three letters. Two statutes with near-identical elements. Two frameworks with the same jargon.
The intervention is counterintuitive. Do not separate them to protect them. Study them side by side, explicitly contrasted, so the difference itself becomes the memory. A comparison table with a column for the discriminating feature does more than three passes over each item alone. This is part of why interleaving beats blocked practice: it forces you to tell similar things apart at the moment of learning rather than at the exam.
The retrieval side
There is a stranger version. Practising some items from a set can make the unpractised items from the same set harder to recall later, an effect known as retrieval-induced forgetting. Bringing one memory forward actively suppresses its competitors.
The implication for studying is uncomfortable but useful: partial review is not neutral. Reviewing half a chapter can leave the other half worse off than not reviewing at all. If you are going to do retrieval practice, cover the set.
See retrieval-induced forgetting for that effect in detail, and active recall for the practice it qualifies.
What this changes about how you study
Space your sessions. Distributed practice reduces interference partly by varying the context each memory is encoded in, giving each one a distinct set of retrieval cues. The spacing effect and spaced repetition.
Vary where and when you learn. Distinct contexts produce distinct cues, which reduces competition between similar items. Context-dependent memory.
Sleep after learning something important. Consolidation happens then, and nothing new arrives to compete. Sleep and memory and memory consolidation.
Do not study two similar subjects back to back. Put something unrelated between them, or a break.
Contrast rather than isolate when two things are genuinely confusable.
Overlearn the things that must survive competition. Extra practice past the point of first success buys resistance to interference. Overlearning.
What it changes about notes
Interference is a strong argument for writing things down, and a specific argument about how.
If similar items are what get lost, then your notes should be built around the distinctions rather than the items. A note that says what X is will be crowded out by the note that says what Y is. A note that says how X differs from Y is doing the work interference makes hard.
It is also a reason to keep atomic notes with precise titles. Three notes with near-identical titles are three competing retrieval cues, and you will reliably open the wrong one. Precision in naming is interference management. See how to name notes.
And it is the clearest case for external memory generally: a written note does not compete with other notes. It sits there while your recall of it decays, which is exactly the division of labour you want.
Decay is not entirely dead
The honest position is that interference explains a great deal and probably not everything. Some forgetting over very long intervals looks more like decay or like the loss of retrieval routes than like active competition, and separating the accounts experimentally is genuinely hard.
For practical purposes it does not matter much which is primary. Both point the same way: memories you never retrieve get harder to reach, and memories with close competitors get harder faster. Retrieval practice and deliberate contrast address both.
More in how memory works, the forgetting curve, and why writing helps memory.