This paper explores the factors influencing analogical transfer in problem solving, focusing on surface and structural similarity. Two experiments were conducted to investigate how analogies are retrieved and used. Experiment 1 demonstrated that spontaneous analogical transfer can occur even with a delay between the presentation of the source and target analogues. Experiment 2 examined the influence of different types of similarity between analogues. The results indicated that both structural and surface features influence the selection of an analogy, with structural features having a greater impact on the ability to use an analogy once its relevance is pointed out. Analogical transfer is a key mechanism that allows human problem solvers to be more flexible than current expert systems in artificial intelligence. Understanding the mechanisms of analogical transfer can provide insights into human cognition and suggest remedies for the brittleness of mechanized problem solvers. Analogical problem solving involves four basic steps: constructing mental representations of the source and target, selecting the source as a potentially relevant analogue, mapping the components of the source and target, and extending the mapping to generate a solution to the target. The second step, selecting a source analogue, is particularly challenging. Computational models of analogy have typically avoided this issue by either explicitly directing the program to compare particular situations or by implementing a psychologically implausible exhaustive search mechanism. Carbonell suggested that retrieval of problem analogies could be facilitated by organizing the database according to similarities in basic problem components, but this approach has not been demonstrated to be effective. The results of Experiment 1 showed that subjects who had previously read a story about a lightbulb problem were more likely to spontaneously apply the solution to a radiation problem. Experiment 2 further demonstrated that both surface and structural similarities influence analogical transfer, with structural features having a greater impact on the ability to use an analogy once its relevance is pointed out. The study also found that structural dissimilarities significantly impair the analogical mapping, while surface dissimilarities have less of an effect. The findings suggest that structural features are more important for analogical transfer than surface features.This paper explores the factors influencing analogical transfer in problem solving, focusing on surface and structural similarity. Two experiments were conducted to investigate how analogies are retrieved and used. Experiment 1 demonstrated that spontaneous analogical transfer can occur even with a delay between the presentation of the source and target analogues. Experiment 2 examined the influence of different types of similarity between analogues. The results indicated that both structural and surface features influence the selection of an analogy, with structural features having a greater impact on the ability to use an analogy once its relevance is pointed out. Analogical transfer is a key mechanism that allows human problem solvers to be more flexible than current expert systems in artificial intelligence. Understanding the mechanisms of analogical transfer can provide insights into human cognition and suggest remedies for the brittleness of mechanized problem solvers. Analogical problem solving involves four basic steps: constructing mental representations of the source and target, selecting the source as a potentially relevant analogue, mapping the components of the source and target, and extending the mapping to generate a solution to the target. The second step, selecting a source analogue, is particularly challenging. Computational models of analogy have typically avoided this issue by either explicitly directing the program to compare particular situations or by implementing a psychologically implausible exhaustive search mechanism. Carbonell suggested that retrieval of problem analogies could be facilitated by organizing the database according to similarities in basic problem components, but this approach has not been demonstrated to be effective. The results of Experiment 1 showed that subjects who had previously read a story about a lightbulb problem were more likely to spontaneously apply the solution to a radiation problem. Experiment 2 further demonstrated that both surface and structural similarities influence analogical transfer, with structural features having a greater impact on the ability to use an analogy once its relevance is pointed out. The study also found that structural dissimilarities significantly impair the analogical mapping, while surface dissimilarities have less of an effect. The findings suggest that structural features are more important for analogical transfer than surface features.