Gambling-like activities are more familiar than traditional gambling in everyday life. Games such as gacha, fukubukuro, and social online games have been linked to problematic behaviors that are similar to those seem in pathological gamblers. However, it is difficult to distinguish gambling from gambling-like activities. In this study, we examined one of the most remarkable differences, which is the existence of minimum rewards, i.e., rewards in “lost trials.” We trained four male Long-Evans Rats to make a choice between a fixed-reward and a variable-reward in a 2-lever Skinner box. According to whether the rat won or lost the trial, selecting a variable-reward resulted in a stimulus that predicted 0 or 4 food pellets in the gambling sessions (GB), and choosing a variable-reward in the gambling-like sessions (semiGB) resulted in a different stimulus that predicted 1 or 3 food pellets. The mean amount of reinforcer was same between fixed-reward choice and variable-reward choice. However, we found that rats with previous GB experience (GB-semiGB) had higher response rates for the variable-reward choice. In contrast, those without previous GB experience (semiGB-GB) did not show higher response rates in the GB sessions.