Abstract
A recurring question in psychology and cognitive science concerns the expression of theories that
are internally consistent and testable. Natural language is unsatisfactory, as theoretical concepts and
mechanisms are not stated with sufficient precision (e.g., Newell et al., 1958; Newell and Simon,
1972; Farrell and Lewandowsky, 2010; Jones et al., 2014). Formal and, in particular, computational
models avoid the problems of vagueness and under-specification by defining the processes and
cognitive mechanisms that occur during a task. They additionally make quantitative and testable
predictions, not only about the link between input and output, but also about fine-grained measures
such as response times and eye movements. Further, such models can perform complex tasks
and, when simulating learning, can use the statistical structure of the environment to help explain
behavior.
This Opinion article briefly reviews the extent to which computational modeling has been
used to develop theories accounting for the learning and use of chunks, schemata, and retrieval
structures. We use the following definitions. A chunk is a “meaningful unit of information built
from smaller pieces of information” (Gobet and Lane, 2012, p. 541), with the qualification that
this information should be of the same kind. A schema is “a cognitive structure for representing
and retrieving classes of typical situations for which a similar response is required of the learner”
(Lane et al., 2000, p. 776). Finally, a retrieval structure is “a set of retrieval cues [that] are organized
in a stable structure” (Ericsson and Kintsch, 1995, p. 216). We should point out that there exist
plenty of definitions for these terms, which is actually an issue for progress in our understanding.
For example, Richman et al. (1991) consider that a retrieval structure is a schema in long-term
memory. Even fuzzier is the concept of a “chunk.” For example, a chunk is a unit of declarative
memory in ACT-R (Anderson et al., 2004) and a unit of procedural memory in Soar (Newell, 1990),
with none of the two meanings corresponding to the definition provided above. For a discussion of
the multiple meanings of this term, see Gobet et al. (in revision).
are internally consistent and testable. Natural language is unsatisfactory, as theoretical concepts and
mechanisms are not stated with sufficient precision (e.g., Newell et al., 1958; Newell and Simon,
1972; Farrell and Lewandowsky, 2010; Jones et al., 2014). Formal and, in particular, computational
models avoid the problems of vagueness and under-specification by defining the processes and
cognitive mechanisms that occur during a task. They additionally make quantitative and testable
predictions, not only about the link between input and output, but also about fine-grained measures
such as response times and eye movements. Further, such models can perform complex tasks
and, when simulating learning, can use the statistical structure of the environment to help explain
behavior.
This Opinion article briefly reviews the extent to which computational modeling has been
used to develop theories accounting for the learning and use of chunks, schemata, and retrieval
structures. We use the following definitions. A chunk is a “meaningful unit of information built
from smaller pieces of information” (Gobet and Lane, 2012, p. 541), with the qualification that
this information should be of the same kind. A schema is “a cognitive structure for representing
and retrieving classes of typical situations for which a similar response is required of the learner”
(Lane et al., 2000, p. 776). Finally, a retrieval structure is “a set of retrieval cues [that] are organized
in a stable structure” (Ericsson and Kintsch, 1995, p. 216). We should point out that there exist
plenty of definitions for these terms, which is actually an issue for progress in our understanding.
For example, Richman et al. (1991) consider that a retrieval structure is a schema in long-term
memory. Even fuzzier is the concept of a “chunk.” For example, a chunk is a unit of declarative
memory in ACT-R (Anderson et al., 2004) and a unit of procedural memory in Soar (Newell, 1990),
with none of the two meanings corresponding to the definition provided above. For a discussion of
the multiple meanings of this term, see Gobet et al. (in revision).
| Original language | English |
|---|---|
| Article number | 1785 |
| Number of pages | 4 |
| Journal | Frontiers in Psychology |
| Volume | 6 |
| DOIs |
|
| Publication status | Published - 24 Nov 2015 |