[ Pobierz całość w formacie PDF ]
produce some of the other effects associated
between studied propositions and inferences.
with inferential memory. This situational or
We decided to adapt the two-strategy model
script information is better retained than the
that we used for Reder (1982) to make the ver-
studied propositional information because it has
batim judgments in this experiment. We as-
received more practice in the past and not be-
sumed that participants were selecting among
cause of different retentive properties. We claim
the following strategies.
1. Retrieval strategy: Try to retrieve a ver-
8
G was not estimated in the model (Table 2) for Anderson
batim trace (e.g., Fig. 2a) to match the sen-
(1974) because accuracy remained at ceiling over the short
tence. Only if this fails go on to retrieve a
period of that experiment and so did not differ between the
two strategies.
propositional trace (e.g., Fig. 2b). If no
358 ANDERSON, BUDIU, AND REDER
FIG. 11. Results from Zimny (1987) and ACT-R predictions. Data are plotted for the two types of judgments (Recog vs
Plaus) and type of sentence (Old, New, and Implausible).
such trace can be retrieved assume the sen- than the G from Reder (1982), which was 34, is
tence was not studied. This strategy will re- interpreted as Zimny s participants placing less
ject inferences and unrelated sentences emphasis on verbatim accuracy than did
since there are no traces of these sentences. Reder s participants on the accuracy of their
It will reject paraphrases if either a mis- plausibility judgments.9 These net utilities can
matching verbatim trace can be retrieved or be converted into probability of choice through
the propositional trace cannot be retrieved.
the Conflict Resolution Equation using the
It will reject verbatim sentences only if nei- same value of the noise parameter t of .05 that
ther the verbatim nor the propositional
was used in the earlier models. We also esti-
trace can be retrieved.
mated a probability .24 of slipping and produc-
ing the wrong response. The overall correlation
2. Inference strategy: Simply determine if
with the data is .956.
the sentence is part of the script. This strat-
As with the Reder model, this model illus-
egy will accept all sentences except novel
trates how participants choice among strategies
unrelated sentences.
is determined by the relative availability of the
We estimate that the shortest delay was 60 s.
memory structures. The verbatim structure is
At this delay, the retrieval strategy will enjoy
the most fragile because it is the most complex
greater success in discriminating verbatim sen-
and the situation referent is most permanent be-
tences (which is the participants task) but will
cause it has been well practiced before the ex-
also take longer to execute since the chunks
periment. There are no inherent differences in
formed to encode the study sentence are weaker
the traces set down in the experiment. It is inter-
than the referent chunks. As time passes, how-
esting to note in Fig. 11 that, according to the
ever, the accuracy advantage of the retrieval
theory, even acceptance of inferences should
strategy disappears as memories decay and
start dropping after 4 days. This trend is only
their latency cost increases just as the re-
slightly apparent in the data but eventually this
trieval strategy lost relative to the inference
would happen as participants come to com-
strategy in the simulation of Reder (1982).
pletely forget the stories that they have studied
Table 4 presents an analysis of the relative util- and so forget the connections of the story to the
ity of these strategies comparable to Table 3 for
9
Reder (1982). The value of G estimated in this
The value of G is really being constrained to produce a
experiment was 10.5. The fact that it is lower 50% strategy mix at the 40-min delay.
THEORY OF SENTENCE MEMORY 359
TABLE 4
predictions of the ACT-R model would depend
on the strategy. In the case of the retrieval strat-
Analysis of Strategy Selection in Zimny (1987)
egy, the model produces a dependence between
Immediate
the acceptance of verbatim sentences and para-
(60 s) 40 min 2 days 4 days
phrases because both will be accepted if there is
a propositional trace and no verbatim trace to
Verbatim Strategy
Accuracy (P) .90 .79 .60 .51
reject the paraphrase. This means that in the ab-
Time (C) 2.71 3.04 3.21 3.21
sence of a verbatim trace either both will be ac-
Utility (PG-C) 4.75 3.78 2.56 2.11
cepted or neither will. However, in the immedi-
Inference Strategy
ate condition of the Zimny experiment, since
Accuracy (P) .67 .67 .67 .67
the propositional trace is almost always present,
Time (C) 2.27 2.42 2.51 2.60
Utility (PG-C) 3.90 3.78 3.70 3.57 [ Pobierz całość w formacie PDF ]
zanotowane.pl doc.pisz.pl pdf.pisz.pl ocenkijessi.opx.pl
produce some of the other effects associated
between studied propositions and inferences.
with inferential memory. This situational or
We decided to adapt the two-strategy model
script information is better retained than the
that we used for Reder (1982) to make the ver-
studied propositional information because it has
batim judgments in this experiment. We as-
received more practice in the past and not be-
sumed that participants were selecting among
cause of different retentive properties. We claim
the following strategies.
1. Retrieval strategy: Try to retrieve a ver-
8
G was not estimated in the model (Table 2) for Anderson
batim trace (e.g., Fig. 2a) to match the sen-
(1974) because accuracy remained at ceiling over the short
tence. Only if this fails go on to retrieve a
period of that experiment and so did not differ between the
two strategies.
propositional trace (e.g., Fig. 2b). If no
358 ANDERSON, BUDIU, AND REDER
FIG. 11. Results from Zimny (1987) and ACT-R predictions. Data are plotted for the two types of judgments (Recog vs
Plaus) and type of sentence (Old, New, and Implausible).
such trace can be retrieved assume the sen- than the G from Reder (1982), which was 34, is
tence was not studied. This strategy will re- interpreted as Zimny s participants placing less
ject inferences and unrelated sentences emphasis on verbatim accuracy than did
since there are no traces of these sentences. Reder s participants on the accuracy of their
It will reject paraphrases if either a mis- plausibility judgments.9 These net utilities can
matching verbatim trace can be retrieved or be converted into probability of choice through
the propositional trace cannot be retrieved.
the Conflict Resolution Equation using the
It will reject verbatim sentences only if nei- same value of the noise parameter t of .05 that
ther the verbatim nor the propositional
was used in the earlier models. We also esti-
trace can be retrieved.
mated a probability .24 of slipping and produc-
ing the wrong response. The overall correlation
2. Inference strategy: Simply determine if
with the data is .956.
the sentence is part of the script. This strat-
As with the Reder model, this model illus-
egy will accept all sentences except novel
trates how participants choice among strategies
unrelated sentences.
is determined by the relative availability of the
We estimate that the shortest delay was 60 s.
memory structures. The verbatim structure is
At this delay, the retrieval strategy will enjoy
the most fragile because it is the most complex
greater success in discriminating verbatim sen-
and the situation referent is most permanent be-
tences (which is the participants task) but will
cause it has been well practiced before the ex-
also take longer to execute since the chunks
periment. There are no inherent differences in
formed to encode the study sentence are weaker
the traces set down in the experiment. It is inter-
than the referent chunks. As time passes, how-
esting to note in Fig. 11 that, according to the
ever, the accuracy advantage of the retrieval
theory, even acceptance of inferences should
strategy disappears as memories decay and
start dropping after 4 days. This trend is only
their latency cost increases just as the re-
slightly apparent in the data but eventually this
trieval strategy lost relative to the inference
would happen as participants come to com-
strategy in the simulation of Reder (1982).
pletely forget the stories that they have studied
Table 4 presents an analysis of the relative util- and so forget the connections of the story to the
ity of these strategies comparable to Table 3 for
9
Reder (1982). The value of G estimated in this
The value of G is really being constrained to produce a
experiment was 10.5. The fact that it is lower 50% strategy mix at the 40-min delay.
THEORY OF SENTENCE MEMORY 359
TABLE 4
predictions of the ACT-R model would depend
on the strategy. In the case of the retrieval strat-
Analysis of Strategy Selection in Zimny (1987)
egy, the model produces a dependence between
Immediate
the acceptance of verbatim sentences and para-
(60 s) 40 min 2 days 4 days
phrases because both will be accepted if there is
a propositional trace and no verbatim trace to
Verbatim Strategy
Accuracy (P) .90 .79 .60 .51
reject the paraphrase. This means that in the ab-
Time (C) 2.71 3.04 3.21 3.21
sence of a verbatim trace either both will be ac-
Utility (PG-C) 4.75 3.78 2.56 2.11
cepted or neither will. However, in the immedi-
Inference Strategy
ate condition of the Zimny experiment, since
Accuracy (P) .67 .67 .67 .67
the propositional trace is almost always present,
Time (C) 2.27 2.42 2.51 2.60
Utility (PG-C) 3.90 3.78 3.70 3.57 [ Pobierz całość w formacie PDF ]