No Evidence of Transfer of Learning Between Problem-Solving Tasks Using Different Transformation Rules

Area of Science:

• Cognitive Psychology
• Artificial Intelligence
• Educational Technology

Background:

• Growing demand for problem-solving skills necessitates efficient training methods.
• Maximizing cross-domain learning transfer is key to efficient problem-solving skill training.
• Current models offer contrasting predictions on the possibility of cross-domain transfer.

Purpose of the Study:

• To investigate the extent of cross-domain transfer in problem-solving skills.
• To compare training outcomes with predictions from pattern-recognition and heuristic search models.
• To explore factors influencing transfer, such as strategy use.

Main Methods:

• Participants underwent four days of training on transformation problem-solving tasks with distinct rule sets.
• Pre- and post-training probes were administered on an untrained set of tasks.
• Participant strategies and decision-making processes were analyzed.

Main Results:

• Training significantly improved solution rate, efficiency, and decision speed on practiced tasks.
• No reliable performance improvements were observed on untrained tasks beyond direct practice effects.
• Explicit deliberation strategies were associated with better decision-making.

Conclusions:

• The study found a lack of significant cross-domain transfer, contradicting heuristic search models.
• Absence of transfer suggests limitations in current models and implications for training methodologies.
• Further research is needed to understand and enhance problem-solving skill transfer.