In the literature, you see people use both the “up-down” method to determine paw withdrawal thresholds (PWTs) and also the ‘frequency’ method. In some papers, both are used, but in general, I find the up-down method to be more common. You can determine a 50% PWT by frequency, as many good papers from the Kuner lab do, but overall, it’s a lot more work and pokes per mouse, whereas up-down is like 5-6 pokes usually.
While in theory I’ve felt that the frequency method is more accurate, I usually use up-down because of efficiency. But the other day, I had an epiphany about why frequency might be superior.
With up-down, there is a requirement for intra-test judgement about responses. You either move up or down based on your judgement that the mouse ‘responded’. And that decision could make a big difference as to the final PWT estimate.
Conversely, the frequency method requires no such judgment. There is a fixed number of applications, regardless of what the mouse does. If you use video, you don’t need to make any judgements on the spot at all. You just poke the mouse in a fixed way and what the mouse does has no bearing on the way the test is performed. This seems to me like it would be much more consistent and easier for testers of all levels to perform. And, the more behavior I do, the more I realize that a binary ‘yes/no’ response is missing a lot of complexity in the withdrawal behaviors. So, better than making a determination during the time of testing, I’d rather just take video of everything and make judgements later, and look for more fine grained behaviors if ned be.
In general, I think the feature I’m picking up on here is tester-dependence vs independence. Having a fixed test that does not require any decision-making on the part of the tester should be superior in terms of accuracy and reproducibility, and we should prefer such methods whenever possible.
What do you think?