Aim

1. To test whether interference during the recall interval affects retention of a verbal item.
2. To test whether the type of rehearsal a participant does affects how much they recall.

Procedure

1. Part 1 used 24 students at Indiana University.
2. On each trial the experimenter spelled a consonant trigram aloud along with a three-digit number.
3. The participant then counted backwards from that number in 3s or 4s, which blocked rehearsal of the letters.
4. Recall was asked for after 3, 6, 9, 12, 15 or 18 seconds, and each participant was tested eight times at each of those six intervals.
5. Part 2 used 48 students, split into a vocal group who repeated the trigram aloud and a silent group who did not.
6. The interval before the counting task began was varied in length, so the vocal group had more or less time in which to repeat the trigram.

Results

1. In part 1, correct recall fell steadily from about 80% at 3 seconds to about 10% at 18 seconds.
2. In part 2, recall improved as the repetition interval grew longer in the vocal group only, with no such improvement in the silent group.

Conclusions

1. More than 90% of trigrams were forgotten after 15 seconds when rehearsal was prevented, so short-term memory has a very short duration without rehearsal.
2. How fast the trigram was lost depended on how much rehearsal happened, which is what part 2 showed.
3. This supports the multi-store model's claim that short-term memory is a separate store with a strictly limited duration.

Strengths and weaknesses

1. A strength is that the procedure was standardised, with the same counting task and the same fixed recall intervals for everyone, so it can be repeated exactly.
2. A second strength is that the counting task held rehearsal out and let only the length of the delay vary, so it isolates the variable and gives clear evidence for the duration of short-term memory.
3. A weakness is that you are rarely asked to hold a meaningless string of consonants in mind, so the task has low ecological validity.
4. A second weakness is that decay over time cannot be separated from displacement by the counting task, so you cannot tell which removed the trigram.

Turning this into marks

1. Give the figures, so about 80% correct at 3 seconds falling to about 10% at 18 seconds.
2. Say the trigram was spelled aloud with a three-digit number, and name the counting back in 3s or 4s.
3. Pin low ecological validity to the meaningless consonant trigrams rather than calling the study unrealistic.