Results for ""
CHREST, "Chunk Hierarchy and REtrieval STructures," is a cognitive architecture that simulates how people think, learns, remember, and solve problems. It was based on chess modelling work and influenced the earlier EPAM model.
Peter C. Lane of the University of Hertfordshire and Fernand Gobet of Brunel University are the creators of CHREST. It is the replacement for Herbert A. Simon and Edward Feigenbaum's EPAM, a cognitive model.
Cognition model
The model brings together low-level and high-level aspects of cognition, such as the mechanisms that keep track of information in short-term memory (e.g., using strategies). It comprises ways to interact with the outside world through perception, short-term memory stores (primarily visual and verbal memory stores), long-term memory stores, and related methods to solve problems. In CHREST, chunks of long-term memory are in short-term memory.
Large corpora of stimuli representative of the domain, such as chess games for simulating chess expertise or child-directed speech for simulating children's language development, have frequently been used in CHREST to model learning. In this regard, simulations performed with CHREST resemble those performed with connectionist models more than conventional symbolic models. Moreover, chunking networks, which reach trees and connect and store knowledge and information acquired, are how CHREST holds its memories, enabling more effective information processing.
Architecture
The architecture includes capacity parameters, such as the visual short-term memory capacity. It is in three chunks (e.g., time to learn, a chunk of time to put information into short-term memory). As a result, it is possible to derive quantitative predictions about human behaviour.
The model considers how individuals interact with things outside of themselves, how their short- and long-term memories are stored, particularly their verbal and visual memories, and how they approach problems. Chunks in CHREST are through the neural perception that involves discrimination and are referenced in short-term memory while stored in long-term memory. Like EPAM, chunks of information learned through cognition are stored and sorted in long-term memory as a "network of nodes" connected by similarities in their contents.
In chess play, chunks are essential "clusters of information that can be used as units of perception," so fragments and parts of chess positions will be to the system as stimuli. The chunks adapt based on recurrent environmental patterns and structures to form cognitive templates, better known as schemas. Templates are mental models of the world that facilitate cognitive organization, memory, behaviour guidance, situational prediction, and general comprehension. In addition, each template has areas where values can be "slotted in," allowing for quicker understanding when presented with information already in the template.
Conclusion
Before de Groot and Simon came up with their ideas and put them into practice, chess experiments and research usually involved showing a chess position to a subject for a short time, usually 5 seconds, and then asking the issue to recreate the position. In this method, the skill level of the subject, the time spent illustrating the point, and the general depth and importance of the point are all standard independent variables. Unfortunately, even though this method has led to many high-level models about memory and cognition in chess play, there aren't many models that go into more detail.
The information provided on this page has been procured through secondary sources. In case you would like to suggest any update, please write to us at support.ai@mail.nasscom.in