NATURE September 26, 1964 v,,.

203

placed before or behind tho screen simply by reversing out the first four from memory. The crucial variable was
the spectacles. the similarity of the eight letters which had to be copied
Suitable light sources are ininiature filamcnt, 12-V, to the four which were being remembered. The letters
100-W tungsten-iodine lamps. The filarr~entsshould be used were taken from two groups I' L M H S X and
oriented vertically, to give maximum horizontal resolution, BCDGI ' T V (Z). ConradQas shown that within each
which is important for giving maximum information of group acoustic coi~fusior~s are common, whereas between
depth through disparity. the groups they are rare. ( 2 does not confusc with B C,
Magnification of the normal z- and y-axes is given ctc., in English; but with its America,n pronunciation it is
simply by the ratio of the distance of the object from the obviously similar to the others and Wickelgren added it.)
lights to the object from the screen. Magnification in Wickelgren's results indicate that when thc letters to be
depth (z-axis) can be controlled by the horizontal separa- copied came from the same group as those being rcmcm-
tion of the lamps. I n practice a rnagr~ificationof a t least bered, memory was poorer than when they carrlo from the
times 10 may be given; but tho apparent size of tho othor group. They are not clear-cut, however, since he
three-dirnensional shadow-image is generally less than had to introduce repetition. To illustrate this: if B C D G
might be expected, because of perceptual size constancy. wero to bc remembered and sin~ilarletters had to be
When tho object appears to be in space before the screen copied, the only four letters available were P T V Z, so
it loolrs sn~allerthan when behind it, though the physical subjects were required to writo each twice. With the 'I("
size is of course identical with either orientation of the sub-group tho problem was even more acute since thcre
polaroids. A further pcrcoptual t:ff(:ct is most marked: were only two spare letters.
when the observer moves his head, the stereo shadow- The experiment reported here was carried out in order
irnage appears to move with him, to slide across the sercen, to confirm Wickclgron's findings. To avoid the problems
and rotate to follow the observer as though continuously already outlined here the subjects were given three letters
aimed a t him. This is because thero is no motion parallax to remerr~ber and six to copy down. The six always
although tho image lies perceptually in three dimensions: consisted of three different letters each repeated once.
this corresponds t o a normal object rotating t o keep the For this experiment the 'B' group was pruned to
sarno aspect to the obscrver though he moves. So he secs B C U G IF.' 8. Forty trials were run, 10 with 'P' lcttcrs
rnovorncnt though his retinal images remain unchanged. to be romembered and ' F ' letters copied; 10 with ' F '
Whon the obsr:rver moves away from the screen, the letters to be remembered and 'U' letters copied; 10 with
shadow-object does not shrink or losc its depth as might 'B' letters to be remembered and 'P' letters copied; and 10
be expected; the rodueed angle of convergence of the eyes with 'B' to be romembered and 'B' copied. Two serics of
evidently re-scales tho retjiial disparity mechanism, so trials were used. The first was prepared by randomizii~g
that a given disparity between the retinal images gives the order of the 40 trials. Tho second was derived from
grzater visual depth. This is indeed fortunate, for the the first by keeping the messages to be renlernbercd
shadow-image appears very similar over a wide range of constant and altering the mclssages to be copied, so chang-
ing if'-F to 1"--B,I"--Bto F-F a,nd so on. Any influence
distance, and so it can be used with effect in n, large lecture
hall for demonstrating suitable objects. They may be of order of prcsontation was in this way controlled. All
moved or touched by tho demonstrator, which makes items were read out by tho oxperimcntcr in time with a
this a techniquo having advantages over photographic metronome beating a t a 2,'sec rate. A pause of 0110 beat
stereo projection. It is also a useful tool for investigating separated tho memory message from the copy messu.ge.
perception of depth while the observer is moving, and can Ten seconds separated the end of one copy message from
form tho basis of simulators for experiments on the the begilmiilg of the next memory messa.ge. Subjects
guiding of aircraft and space vehicles. were tested in two groups of 17. They wrote out all
This technique was devised for experiments supported messages on a specially prepared form. An invigilator
by U.S. Air Force grant AP-EOAR 63-93, monitored checked that they put down the copy messego before the
by the European Office, Office of Aerospace R'esearch. memory one.
It. L. GREGORY Performance was scored by counting the number of
letters in caeh memory message which were correctly
Psychological Laboratory, recalled in both substance and position. The results
University of Cambridge. showed a mean score of 59.4 per cent when the copicd
message came from a different group of letters to the
memory message but only 38.7 per cent when the copied
Retroactive Interference in Short-term message came from the same group. The individual
Memory records of 31 of the 34 showed this effcct; its statistical
significance is, therefore, extremely high (P < 0.001).
INinvestigating short-term memory (a term generally Thus Wickelgren's observation is confirmed and we must
used to cover the period of from one to sixty seconds) an conclude that the principle of retroactive interference does
important cjuestion is whether or not retention is in- apply t o short-term memory. The form of the similarity
fluenced by factors known t o be important in long-term which affects performance, however, would secbmto differ
memory. One such factor is retroactive interference. from that which influences interference in long-term
Broadbcntl has asserted that retention depends on the memory. Meaningful similarity is unimportant ovel.
duration of the delay rather than the nature of any these short periods, but similar-sounding items cause
activity during the delay. But he rests his case on interference. Presumably the basis of coding for normal
evidence that retention is impaired by increasing the
verbal learning with repeated presentations and with
delay while kocping the activity as constant as possible. opportunity for rehearsal differs from that used for
Direct evidence that variation of tho intervening activity short-term storage following a single presentation.
is ineffective has been lacking, but Baddeloy and Dale2
have shown that in minimal paired-associate learning3 H. C. A. DALE
meaningful similarity fails to affect performance sigr~ifi- Applied Psychology Research Unit,
cantly, and this supports Broadbent's contention. Medical Research Council,
15 Chaucer Road,
Recently Wickelgren4has produced cvidence suggesting
that acoustic similarity does affect retroactive interference Cambridge.
in short-term memory. His technique was t o present ' Broadbent, D. E., J. Verb. Learning and Verb. Behavior, 2, 34 (1963).
Baddeley, A. T)., and Dale, H. C. A. (in preparation).
rapidly (2/sec rate) four letters which had to be remem- Peterson, C. R., and Peterson, M. J., J. Ezp. Psychol., 63, 521 (1962).
berod, then eight which wero to be copied down. Imrned- Wickelgren, W. A. (personal communication).
iately the eight had been copied the subjccts had t o writo Conrad, R., Brit. J. Psychol., 55, 75 (1964).