This week we are going to have a closer look at experiments. As explained in detail in Beth’s book, by manipulating one or a few variables and by holding constant other variables, experiments allow us to draw causal inferences. If the experiment has a between-subjects design, random allocation to groups is another key characteristic of experiments.
Experiments are all about explaining an observed effect on the dependent variable (DV) as a result of the manipulation of the independent variable (IV). If the observed effect is actually the result of manipulating the IV, the experiment is referred to as internally valid (meaning that it appears there are no alternative explanations for the observed effect). Threats to internal validity are a key issue for experiments and are discussed in detail in Chapter 11 in Beth’s book.
Our focus in the labs will be on a number of experiments that measure response times (RTs)9 and accuracy. RTs have a number of advantages that make them suitable for our lab classes: they can be easily acquired using standard keyboards, they are objective and they are sufficiently accurate for our purposes.
It is important to realise why RTs and accuracies are of interest to us: They provide us with information about the cognitive and neuronal mechanisms taking place in your minds and brains. Any processing going on in our brains takes time. And more complicated things take more time (and typically increase error rates)! Which is great for us as psychologists as this gives us the opportunity to study how our mind and brain work by studying RTs and error rates!
What is good for us is that there are a number of experimental paradigms based on measurements of RT and accuracy that result in highly consistent effects if participants follow the instructions (typically, this involves responding as quickly and accurately as they can). Thus, at least on a group level, these effects are highly replicable (see Chapter 14 in Beth’s book for an in-depth discussion of replication). In addition, these paradigms have the advantage that they still work even if the participant is aware of what is being manipulated (again, as long as they do their best to follow the instructions).10
Let’s take a look at what is the classic example of the idea that adding extra processing demands slow us down: The comparison of simple and choice reaction time. This idea was introduced more than 150 years ago in one of the most influential publications in the history of psychology:
The idea occurred to me to interpose into the process of the physiological time [i.e., simple reaction time, J.D.] some new components of mental action. If I investigated how much this would lengthen the physiological time, this would, I judged, reveal the time required for the interposed term.
F. C. Donders (1869), On the speed of mental processes
Below, you will have the opportunity to find out how long it takes you to make a very simple decision! As mentioned above, please do your best to respond as quickly and as accurately as possible.
Try out a simple reaction time task. There are 24 trials overall. Once finished, you will be presented with your mean RT for this task.
If you would like to rerun the task, please reload the page.
Try out a choice reaction time task. Again, there will be 24 trials overall. You must get at least 21 out of 24 trials correct to obtain your mean RT for this task.
Please enter your data below (reload the page if the app says that you have been disconnected).
Note that the cost of making a choice is typically between 75 and 200 ms. Given that RTs in a simple reaction time task are typically between 200 and 350 ms, this represents a substantial cost. Let me know if you consistently have a choice cost of less than 50 ms despite doing your best to respond as quickly as possible in the simple reaction time task. If that’s okay with you, I might ask you to complete a couple of additional tasks.
The terms response time and reaction time are often used synonymously and refer to the length of time from stimulus onset (e.g., a picture on a screen) to movement offset (e.g., pressing down a key with a finger). Sometimes response time is defined as the sum of reaction time (i.e., time from stimulus onset to movement onset) plus movement time (i.e., time from movement onset to movement offset). Following this definition, in the above example reaction time would be measured from picture onset up to the point when the finger starts moving. Movement time on the other hand would be the length of time from when the finger starts moving to when the key is fully depressed. For simple responses (such as key presses), this distinction doesn’t usually matter. For more complex responses (such as, say, pointing responses involving whole arm movements), separating reaction and movement time can be of interest.↩︎
Note that this is not the case with many of the interesting studies mentioned on the forum. Many of those manipulations only make sense if the participant is not aware of what is being manipulated.↩︎