QUESTION:
Does the problem of measuring the ability of a subject (e.g. an other than human animal) to perform a task has a name?
Are there standard methodologies to do so ethically, especially when checking that a subject is unable to perform the task?
EXAMPLES:
In order to validate the cognitive ability of a subject to, say, discriminate the largest between two amounts,
- one can
- A. present the subject with the choice between two food quantities, reward them with their choice, and measure how often they do choose the largest one; or
- B. present the subject with the choice between two quantity representations, reward them when they choose the largest one, and measure how often they do so.
- If the subject is indeed able to perform the task, both methodologies are mostly equivalent, but if the subject is NOT able to perform the task, each experiment will be a frustrating event, and forcing the subject to repeat the experiment will neither be ethical nor productive, which lead to consider the third type of experimental protocol:
- C. present the subject with the choice between two quantity representations, reward them whatever the choice but with a preferred prize when they choose the largest one, and measure how often they do so.
- one can
In order to validate the sensory ability of a subject to distinguish, say, blue and green colors on a screen,
- one could
- A. display challenges that the subject is known to be able to solve when displayed in other way, but using blue on a green background (or the reverse), and check if the subject can still solve the challenge; or (more complicated)
- B. display (e.g. cognitive) challenges which can be interpreted at two levels depending of the sensory abilities of the subject (e.g. one value is blue points on green background, the other red points on green background) , rewarding correct answers at one sensory level (e.g. selecting the largest value if one cannot distinguish blue from green), but rewarding even more correct answers at the other sensory level (e.g. selecting the largest value if one CAN distinguish blue from green).
- I am thinking about implementing the (more complicated to implement, and even more so to analyze and to explain to others), but I fear that I am reinventing the wheel, and maybe making it more complicated than necessary. I wonder if the problem was previously considered (probably in another setting!), under which name, and if there are there any classical solutions (and what are the criteria to compare potential solutions to such a problem)?
- one could
SOME CONTEXT:
I programmed a very simple software requesting subjects to choose the maximal value out of 2,3,4 or 5 values, represented in various modes (and used it with success to study the some discriminatory abilities in Monk Parakeets 2). I programmed it so that one could change the colors, first with the objective to find the contrast and color settings most adequate for each subject. I later thought that I could use it to study the limits of the subject's perceptions, but I am uncomfortable about requesting subjects to solve instances that they are unable to solve. I could progressively alter the colors and contrast, and study how the correction rates vary with the parameter altered, but it seems both slow and a bit "cruel". I felt comfortable doing so with increasing progressively the number of values to pick from and the range of values from which the displayed values are shown because it was also pedagogical, but doing so with sensory limits feels different to me.
