Documente Academic
Documente Profesional
Documente Cultură
JOHN A. ROBINSON
University of Louisirille
The Underwood and Keppel (1963) coding task was analyzed into three com-
ponents: encoding, solution-word recall, and decoding. It was argued that
5s could be expected to formulate and use letter-order rules to mediate
coding. The principal concern of the paper was with the effects of intralist
rule order (IRO) on the learning of a two-rule list of codeable trigrams.
IRO was varied by presenting rule-associated trigrams in successive blocks
or in successive alternations. Four experiments were performed. The
results indicated that each task component was significantly affected by IRO
though the basis of the effects varied among the three components.
Underwood and Keppel (1963) devised However, lists could be constructed where
a task for exploring the role of coding subsets of trigrams are associated with dif-
processes in verbal learning. They had 5s ferent LORs.
learn a list of trigrams which could be en- The effect of varying the number of LORs
coded into words (e.g., UBT: TUB; BUT) in a list was investigated by Underwood
and required some 5"s to reproduce the tri- and Erlebacher (1965). They report that
grams during recall, but permitted others to a two-rule list is significantly harder than
reproduce the constituent letters in any a one-rule list but that increasing the num-
order. List learning progressed more ber of LORs to four does not appreciably
rapidly under the latter condition. It would increase task difficulty. Such a conclusion
seem that when literal recall is required, implies that 5" attempted to use the coding
encoding retards learning. A plausible ex- scheme built into the materials by the in-
planation for this effect is that decoding vestigators. However, in the experiments
the attempt to reconstruct the trigram let- cited thus far encoding has always been
ter order from the word letter orderis dif- optional. Consequently, though error analy-
ficult, especially when, as was the case for ses provide suggestive evidence for the
the Underwood and Keppel (1963) mate- occurrence of encoding and decoding,
rials, there is no rule by which the trans- whether 5"s attempted to formulate LORs
formation can be accomplished for all of the for either of these transformations has re-
trigrams in the list. mained a matter of conjecture. In a recent
It is apparent that decoding could be investigation (Robinson, 1967) it was
accomplished more systematically if the ma- shown that 5"s will discover and use an LOR
terials were such that letter-order rules for encoding and that predictable disruptions
could be formulated to mediate this trans- of performance occur when LORs changed.
formation. A letter-order rule (LOR) In view of this result it was decided that
would be any statement of the letter moves rule-based coding merited more detailed con-
required to produce the word from the tri- sideration.
gram, or conversely, to reproduce the tri- The particular aspect of rule-based coding
gram from its respective solution word. In chosen for study was that of intralist rule
the simplest case all of the trigrams in a order (IRO). Consider a list of codeable
list would be subject to the same LOR. trigrams devised so that two different LORs
l
This research was sponsored by the Air Force may be used for coding, each for half of
Office of Scientific Research, Office of Aerospace the items. Under these conditions the order
Research, United States Air Force, under Grant of item presentation is concurrently an order
AF-AFOSR-1008-66. Grateful acknowledgment
is made of the assistance of Barry Rabin in the of rule presentation. That is, if S attempts
running of 5s and the analysis of data. to formulate LORs, each item provides an
INTRALIST RULES AND CODING DURING LEARNING 7
opportunity not only to study a particular affected by IRO. Encoding requires time
trigram but to attempt to determine which and the amount of time needed covaries with
LORs are represented in the list as well. a variety of factors (cf. Johnson, 1966).
Thus, when the order of trigram presenta- Thus, for a list of trigram anagrams where
tion is randomized, rule order is also ran- several of these factors are freely varied,
domized. It may be safely presumed that solution time would not be expected to be
such an arrangement is not the most efficient related systematically to the order of item
way for rule learning to progress. The presentation. If, however, 5" has discovered
question of interest then is whether other a rule by which he can accomplish the de-
nonrandom arrangements of rule-associated sired transformation for a whole class of
trigrams differentially affect the rate at instances, then encoding time should exhibit
which a list of codeable trigrams is learned. a systematic relationship to the order in
In order to evaluate the effect of varying which the rule-governed sets are presented.
IRO two types of lists were constructed, a Specifically, when same-rule trigrams are
blocked (BLK) list and an alternating presented in a block, encoding time should
(ALT) list. In the former type of list all decline rapidly but increase again when a
of the trigrams associated with one LOR different rule set is encountered. In con-
are presented in succession and are, in turn, trast, when trigrams associated with dif-
followed by all of those associated with the ferent rules are presented in alternation,
other LOR. In the latter type of list the solution time should, on the average, be
LOR-associated trigrams are presented in greater at most list positions than it is for a
successive alternations. corresponding BLK list and, further, should
IRO should directly influence ease of de- exhibit a saccadic pattern reflecting the
coding. Specifically, a BLK list should be "switching" character of encoding.
easier to decode than an ALT list. A BLK With regard to a third component of the
list could facilitate decoding in any of the tasksolution-word recallthere seems to
following ways: (a) The critical letter-order be no reason for supposing that IRO would
cues which define the two LORs should be differentially affect this in any way provided
more readily discriminated with successive that an equal number of trigrams have been
presentation of same-rule trigrams, i.e., rule encoded with each type of list. Solution-
formulation should be facilitated. (6) Suc- word recall could be influenced indirectly,
cessful decoding presumably involves recon- however, if as a function of IRO, there were
structing trigram letter order from solution- differences between lists in number of tri-
word letter order thus requiring the asso- grams encoded. An implication of the pre-
ciation of LORs with appropriate item sub- ceding discussion of the possible effects of
sets. A BLK presentation should define IRO on encoding is that with a short ex-
these subsets more easily and should permit posure interval more items would be success-
more effective association of rules and items. fully encoded with a BLK list than with an
(c) The LORs could be used to group the ALT list. Thus, it could be expected that
respective trigrams in some fashion, i.e., the when the exposure interval is brief, IRO
LORs could provide an organizational would influence word recall, but when the
schema for recall. It should be easier to exposure interval is sufficient for complete
exploit the organizational aspects of LORs encoding, IRO would have no effect on word
when rule-associated items are themselves recall.
grouped. The conclusion was then that IRO may
There is no a priori reason to suppose affect each of the three components of the
that decoding is the only component of the taskencoding, word recall, and decoding
task which would be affected by IRO. In- and that the cumulative effects of IRO
deed, if Ss did attempt to formulate and use would be such that list acquisition would
appropriate LORs, it could be supposed that progress more rapidly for a BLK list than
encoding as well as decoding would be for an ALT list.
JOHN A. ROBINSON
words and decoding rules can be associated. observations of Underwood and Keppel (1963)
Finally, general organizational influences may suggest, decoding is the major source of diffi-
also be considered for that would seem to be culty in this type of learning.
the most satisfactory way to account for the
enhanced solution-word recall that was observed REFERENCES
for a BLK list, as well as the high incidence BOUSFIELD, A. K., & BOUSFIELD, W. A. Measure-
of clustering observed for both word and tri- ment of clustering and of sequential constancies in
gram recall. repeated free recall. Psychological Reports,
2. A consistent picture of the relative con- 1966, 19, 935-942.
tribution of each component to performance has JOHNSON, D. M. Solution of anagrams. Psycho-
emerged from these experiments, viz., in order logical Bulletin, 1966, 66, 371-384.
ROBINSON, J. A. Rule learning and rule inter-
of diminishing influence, decoding > word re- ference in trigram encoding. Psychological Re-
call > encoding. This ordering can be further ports, 1967, 21, 921-927.
interpreted in terms of the rate at which each UNDERWOOD, B. J., & ERLEBACHER, A. H. Studies
component approaches asymptotic performance. of coding in verbal learning. .Psychological
For example, in the case of a BLK list encod- Monographs, 196S, 79(13, Whole No. 606).
ing is virtually complete after Trial 1, word UNDERWOOD, B. J., & KEPPEL, G. Coding processes
recall is nearly perfect after Trial 3, but tri- in verbal learning. Journal of Verbal Learning
gram recall is still significantly short of asymp- and Verbal Behavior, 1963, 1, 250-257.
tote by the end of practice. Hence, as the (Received December 31,1967)