Using progressive ratio schedules to evaluate tokens as generalized conditioned reinforcers

Russell et al. (2018) wanted to know if tokens are really strong rewards on their own. Tokens are small items, like plastic chips, that kids trade in later for a prize. The big question was simple: do tokens stay strong even when the prize behind them loses its pull?

The team worked with three children who had autism and developmental delays. They set up a test where the kids had to work harder and harder for the same reward. Sometimes the reward was tokens. Sometimes it was a snack. Sometimes it was a toy or leisure item.

Here is the clever part. Right before some sessions, the researchers let the kids eat as many of the snacks as they wanted. That way, the snack was not exciting anymore. Then they watched to see if the tokens still worked.

For two of the three kids, tokens beat snacks. The kids kept working for tokens even when the task got hard. That is a big deal, because it means tokens had real pulling power of their own.

The tokens also passed the tough test. When the kids were full of snacks, they stopped working hard for snacks, just like you would expect. But they kept working for tokens. The tokens did not fall apart with the snacks.

That is what behavior analysts mean when they call a reward a "generalized" reward. It keeps working even when one specific prize stops being fun, because the tokens can still be traded for other things.

At first, this paper looks like it argues with Butler et al. (2021). That study followed kids for a whole year and found that snacks were actually the most steady, reliable reward of all. Leisure items and social attention came and went, but snacks kept their pull month after month. So you would expect snacks to be the workhorse of any token system.

But Russell et al. (2018) saw snacks crash fast. So who is right? Both, actually. The trick is in how each study set things up. Butler et al. (2021) checked snack strength under normal conditions, across months. Russell et al. (2018) let kids eat their fill of snacks right before the test. Snacks are steady over weeks and months. They are fragile in the moment, right after a kid gets a bunch of them. Tokens sidestep that problem because their value is spread across many prizes, not just one.

This lines up with a long line of work in the field. Dixon et al. (2016) showed the same basic idea with adults who gamble: if you change what someone wants in the moment, you change how much a reward is worth. Catania (2021) makes the theory point that rewards shape whole groups of behaviors, not just one exact action, which fits the idea of a token standing in for a class of prizes. Baum (2025) explains why kids give up faster when they have to work very hard for very little, and that is exactly what progressive ratio tests are built to catch. Regnier et al. (2022) pulls all this together in a review and treats studies like this one as key building blocks for planning strong token economies.

If you run token boards, this is your green light to trust them even on days when the usual snack has lost its spark. But do not skip the check. Before you roll out a new token system, spend a few minutes testing your tokens against your strongest snack. If the kid works harder for tokens, you have a strong, all-purpose reward. If not, fix the trade-in menu first. The same idea scales up in adult settings too, as May et al. (2020) showed with a lottery ticket system that kept adults moving during tough workouts.