Understanding Reaction Time Studies in Cognitive Psychology: History and Methods

Hello and welcome to the course basics of experimental design for cognitive psychology. I am Arura from the

department of cognitive science at Kpur. This is the seventh week of the course and we are discussing reaction time

studies. Before we go forward, let's talk a little bit

about where do these studies started. In the previous lecture, I talked to you about the logic and the goals of the

experimental of the reaction time studies. But where does this start? So, systematic observation and comparison of

response times as a form of scientific inquiry is not new. It goes back to the 19th century in a field that may sort of

appear hardly related to psychology. For example, astronomers uh at the time used what is called the eye and the ear

method to uh you know for timing and recording uh stellar transits. All right. So what they were doing was they

was basically looking at the movement of star in the field of in the field of the telescope and they listened to and

counted the click clicks of a clock and then recorded the exact time when a star reaches a particular position in the

field of the telescope. So from point A to point B they would calculate the time it takes a star to reach from point A to

point B and that is how you they used to time it. So because of this importance of timing uh and timing accuracy in uh

astronomy a great deal of research was done in the 19th century to understand individual differences in observation

and measurement accuracy. So this is basically uh a you know broad background of where measurement of timing sort of

uh started becoming important. Now some of this research was picked up by psychologists as well. For example,

German physiologist Herman Helmolds studied the speed of the sensory nerve conduction by measuring how fast humans

responded to stimula applied to different parts of the body. So basically there and depending upon how

uh far that part of the body was from the brain. So for example how fast you would respond when somebody pricks at

your toe versus how fast you would respond when somebody pricks at the ear given the distance of the brain uh you

know from that body part. However it was Dutch physiologist uh Francis Daers who started a new era of reaction time

research. Now Dundas was interested in measuring the speed of the mental processes just like we were describing

or uh talking about in the previous lecture. So he was more interested in understanding the speed of the mental

processes rather than the speed of physical responses. All right. So he pointed out that if a simple response

task is created just like the one Helm was using such as responding uh to a stimulus on the skin it could consist of

a number of steps only a part of which would basically be uh the timing of the mental process. So for example when you

are giving uh somebody a task and you are expecting them to respond only part of that response time is mental process

rest would be physical sort of uh you know uh receiving the motor signal and executing that motor signal and people

can be fast and slow in that. So if the duration from the stimulus to response was considered as a

physiological time, the total amount of time for individuals to physically respond to stimulus and the time taken

up by the mental process was only a part of the time. So whatever response time is there, it basically according to

Dondas it was two things. It was first the you know time of mental process plus time of physically responding. What he

wanted to do was he wanted to create situations so that he could isolate the time of the mental process there. All

right. So given that there was this very brief response times that were getting measured in such task for example these

could be as short as 17th of a second. It became difficult it I mean he figured out that this was difficult to determine

how much time it would have taken the brain to respond to uh this process brain to perform this particular

computation. So in order to measure the speed of these mental processes he developed a

new approach. This approach is referred to as the subtraction method and it is very common and it is now used not only

in reaction time research but a lot of other kinds of research as well that we will visit in this week and the next

one. Now uh this subtraction method basically involved the comparison of responses obtained in the different

tasks. He described this uh his reasoning for uh his uh for using this approach uh in so many words. Let's see.

He says the idea occurred to me to interpose to into the process of physiological time some new components

of mental action. So you basically said okay if a particular mental action takes so much time let me add a step or

subtract a step from this mental process. If I could investigate how much this would lengthen this physiological

time this would I just reveal the time required for the interposed term. So whatever step that I have added in this

particular sequence. So how do we do this or how did uh Dondas propose? He says we can design two tasks such that

task A requires all the steps that are required in task B and one additional step. All right. So under such a

circumstance what will you do is you will basically expect that it should take longer to complete task A than task

B. one can infer how long it takes to complete that specific additional amendmental process that we have added

in task A by subtracting the time taken to task complete uh to complete task B from the time it was it has taken to

complete task A. So now when you do task B minus task A you get the time for that specific extra computation that you have

added. All right. Following this reasoning, he used three procedures. In the A method, it was

called a simple reaction time task. Here, a participant was asked to respond to a pre-desated stimulus, such as

pressing a button as soon as a light was turned on. So, this is a simple reaction time task. In the B method, he used what

is called a choice reaction time task. A participant was asked to respond differently to a different to different

stimulus. So for example, use one hand to respond to a red light and use another hand to respond to a white

light. So in the first one, as soon as you see a light, you press a button. In the second one, you basically see which

light has turned on. And if red light has come, use this hand. If white light has come, use this hand. Now you can see

in A there is just a single step. In B there is a decision that is also involved. Which light has come up? I

evaluate and then I choose my appropriate hand to respond. So he argued that the B method involved all

mental processes required by the A method plus two additional processes. What are those? The distinguishing

between the two stimuli uh or stimulus discrimination that in this case which is the color of the right of the light

it is red color or white color and deciding how to response which hand am I going to use based on which light has

come up. So two extra steps in B as compared to A. So if the participants reaction time in

performing the two tasks A simple reaction time and B choice reaction time were compared, they would be expected to

take longer in performing a choice reaction time task uh than a simple reaction time task. And whatever extra

time uh you know they have taken could be inferred as the time for the two extra steps that were involved stimulus

discrimination and response selection. Now he said okay let's come up with another method. So he developed a third

procedure the C method which is sometimes referred to as the selective reaction time task. All right. Now in

selective reaction time task what would what would happen is that you could add this to differentiate the time between

the two steps. So what are the two steps? Similar discrimination and response selection. So let's see what

the C method is. In C method, the participant is asked to respond to only the light of one color by ignoring the

light uh you know light of other color. So for example, respond only when you see a red light. Don't respond when you

see a white light. So you can see this C method actually quietly goes in between A and B. All right. So this procedure

basically requires the participant to dis discriminate between two stimula the two colors of the light. But it does not

require response selection because a participant has to respond to one kind of light only. As soon as they uh

decipher that okay which kind of task is being which kind of light has come on they just have to know which to respond

to or otherwise not. So response selection that step is gone from here. For this reason it could be predicted

that it would take less time to complete a selective reaction time task as compared to the choice reaction time

task. So now you have a gradation. You have a simple reaction time task. You have a selective reaction time task and

you have a choice reaction time task. This will take the least amount of time. This will take the second most amount of

time and the choice reaction time task because it has two steps will take the most amount of time.

So consequently what they did was by comparing the times taken in performing these two tasks, one was able to isolate

the time for response choice. So response selection, how much time does response selection take? as this was the

mental process that was involved in the B method but it was not in the A method and it was the reason that response

times would increase in the order of A and then C and then B. That is basically how these tasks were set up and you can

see they are set up so neatly that you can actually compare when you compare A and uh B uh you basically get the uh you

know time difference in selection of responses and you compare B and C you basically get the discrimination of

light. So you can very well arrange this task and know by subtraction of what from what you will get the time taken

for a given process. So this is basically from Sternberg. This is just an example of what kind of task you're

doing. So this is Da's uh setup for 18 uh for creating reaction time experiment. Simple reaction time task,

choice reaction time task and selectious reaction time task. This is basically how this is working. So A is light

flashes, press the key. We perceive the left light then decide which button to basically press left key right key and

so on. Now Daisers and his students carried out a large number of similar experiments and they compared reaction

times for the three kinds of tasks. The results generally confirmed their predictions. For example, in a series of

bubble repetition uh experiments three conditions were included. For example, repeating a wall as soon as possible

while knowing which wall was to going to be presented. So a b a e i o u something like that. Repeating a vowel without

knowing which vowel has to be presented. So knowing which vowel has been presented and producing it. C uh

repeating a pre-desated vowel when the stimula might include different vowels as well. So this is the C method. So you

have now A, B and C. He observed that an average of around 201 mills, 284 millconds and 237 millconds were taken

for the three combinations respectively. Based on these results, uh they were able to infer that the mental process

responsible for recognizing a vowel took about 36 milliseconds and the mental process of uh making the choice of uh

response took around 47 millconds. So you do either C minus A which is uh you know selective minus uh simple or you do

B minus C you get the uh time for the specific step. Darn considers the findings that he found and his students

the first determination of the duration of well-defined neural responses. So this is novel and this is sort of you

know a very specialized finding that Daers and his group is able to uh demonstrate.

Now with a discovery of this subtraction method reaction time research could now flourish during the second half of the c

of the 19th century and uh people like William W and his associates at lizig laboratory published a large number of

reaction times uh studies related to a variety of mental processes. So remember William Bund was the one who formed the

first lab you know laboratory in psychology in the year 1879 and he was partly inspired from the reaction time

research that was conducted or the protocols that were developed by Francis Daers in the 1800s.

Many other leading psychologists of the time also worked on mental uh chronometry. For example, cattle's

extensive work on the time taken by cerebral uh operations and Baldwin's uh comparison of sensory and motor reaction

times were all inspired from uh you know the kind of protocols that uh Francis Daers was developing.

However, around the uh turn of the 20th century, reaction time research declines a a little bit as a result of the re

rejection of the subtraction method. Some people started doubting the subtraction method and says it they are

not uh you know it is not creating uh correct computations and therefore uh you know for a long time the method was

criticized. Let's look at why it was criticized. First was that the reactant results obtained with the subtraction

method had a great deal of variations between individuals and between laboratories which basically made it

hard to compare and you know confirm a particular finding because remember at this time people did not know a lot

about control and there there were a lot of var variation probably across conditions and laboratories. The second

was the other had to do with this assumption of pure insertion. Now you see uh simple reaction time to selective

reaction time to choice reaction time. If we are con considering the subtraction as the true indication of

what reaction times uh you know might tell us we are basically going with this assumption of pure insertion. What is

this assumption of pure insertion? The assumption of pure insertion basically is that adding one step or adding an

extra step to the process will not affect the overall thing but it will affect just add just that amount of time

taken uh you know you utilized for that particular step. Actually some people said that when you make a task more

difficult it overall it increases the overall time and not just the time taken to uh you know perform that additive

step and that basically became a very important critique for reaction time research and which is why it was

discontinued for a long time starting the earlier 19th century. Now uh around 1950s reaction time

research was revived again with the development of computer technology and with the rise of information processing

theory of cognition. So around 1950s you see that the information processing metaphor for cognition takes place where

we start talking about inputs and computations and representations and that is where people found that okay

reaction time research is one way it is one tool to basically again delineate steps in information processing for a

given stimulus and that is again around the time where reaction time research started taking up and a lot of uh you

know experiments which were had to do with evaluating word similar face stimuly pictures and so on started

being used. So this former uh you know reaction time research it provided a means for uh

measuring reaction times in uh a more accurate and reliable way. So now they started using computers. Computers had

more precise timing. They had a better way of presenting stimulus. They had a better way of recording response times.

And in that a bunch of the problems that were coming uh in the 1800s late 1800s early 1900s were handled by the presence

of good computers more accurate presentation and measurement durations. Also the latter the information

processing theory as a metaphor as a framework it provided a very nice conceptual framework that validated and

that necessitated the examination of cognitive processes h as happening in discrete stages in some sense confirming

da's initial ideas about how mental processes really work. Hick uh in 1952 and Heyman in 1953

according to Blackman basically uh led uh you know the way in the survival of the reaction time approach to cognition

but Sternberg's work was one of the ones that basically provided the best illustration of how valid reaction time

research was and it basically uh you know overcame the earlier criticisms about this. So let's let's look at that

in in some detail. Now, Sternberg observes uh endorsed uh Da's view that a cognitive task consists of several

discrete stages and reaction times uh could be used to identify or isolate these stages. All right. He adopted

slightly different approach to da let us see instead of comparing the reaction times to complete tasks involving

different stages always used the same experimental task that involved the same number of stages.

He said we'll re use the same uh stages. We compare simple reaction time and selective reaction time to examine

stimulus discrimination. Let's see how he did it. What he did to identify a given stage was to manipulate a factor

that might apply to only one of the stages. So he said okay I know I'll not use different tasks but I'll use uh you

know some kind of manipulation of a factor within that task. Let us see how that works. Now by observing how this

factor would affect an individual's reaction time, he could infer what stage was involved in a given task and how

long it took the participant to complete that method. So he now basically used what is called the additive factor

method. Da was subtraction. This is an additive factor method. Let us see how this differs

now because there was now no need to add an additional stage to study this approach. This approach effectively

overcame the problem that was uh you know uh attached to this uh you know the idea of pure insertion. More

specifically, Sternberg adopted an item recognition task to study the cognitive processes involved in memory retrieval.

So we'll see what he does. It's a very simple task. The participant had to decide whether a digit presented to them

visually was one of the digits that they had seen earlier as well. Okay. The participant's reaction time was measured

as the duration between the onset of this digit and the time when the response was given. So this digit comes

and the participant is to tell whether it was presented earlier or not. So when the digit comes and the participant

gives the response that is the measurement of the reaction time that they are going with.

Now Sternberg postulated that three stages are involved in this item recognition task. First is stimulus

encoding. So whatever digit is presented is to be encoded. It has to be compared mentally. Okay, this digit that is

presented on screen is three or it is four or it is seven something like that. Uh and then comparing this digit

whatever the input is there with the previously seen digit. So for example, previously the person might have seen uh

3 9 and 8. Now you have to compare okay uh and basically then come up with a decision that okay this digit was

presented earlier or this digit was not presented earlier. Okay, the encoding part, the comparison with previous uh

presented digits and the decision and response. Okay, yes, it was presented or no, it was not presented.

Now, in one of his experiments, in one of Sternberg's experiment, the critical manipulation was the number of digits to

be remembered. Okay, so he would be giving them uh some digits earlier before they go go into this experiment.

Sometimes the digits were one, sometimes they were two, sometimes they were up to six. All right. Now the experimental

task was experimental task that he was measuring reaction times in was identical for the participants in these

six conditions. So the task is exactly the same. We have to see whether this digit was presented earlier or not.

Okay. They had to decide. Yeah. So in this design the only difference between the six conditions is how many digits

were presented earlier. So for example if I have shown three here and I presented three here you have to make

one match. If I presented three here and I have presented four here, you have to still make one match. You'll get the

answer no and you'll quickly press. If I presented three and five here and I presented four here, then you have to

compare with three and compare with five. Both times you return with a no answer and you give a different

response. Okay. So in this design, the only difference across the six conditions is how many comparisons had

to be made before a positive or a negative decision can be arrived at. Now the manipulation of the set size the

previously presented digits conceivably only affected the comparison stage because everything else the decision

response making etc will stay the same. This approach allowed Sternberg to uh determine how much time mental

comparison took by comparing the differences in these different conditions. By comparing the differences

in reaction time in these different conditions and because it is the same task almost this determination was not

dependent on pure insertion assumption. Okay, because you're not adding an extra step. It is the same number of steps but

the mental uh you know comparison has sort of increased. you are comparing with one set size or two set size or

three set size worth of digits. This experiment showed that the size of the set how many uh you had to go and

compare with and the participants reaction time actually had a linear relation. So if you're comparing with

one versus comparing with four versus comparing with six, the reaction times increased accordingly. Okay. So the

response latencies actually were shown to increase by 38 milliseconds with each increasing digit to be remembered. All

right. So assuming that the individuals had to scan their immediate memory for responding to the stimulus, these

results actually offered a very good indication of the speed at which we scan our memory.

So this is this you have shown uh four digits ear four alphabets earlier and then you have shown L. Now you have to

compare this L with all four of them and it basically will take you a certain amount of time. Let's say if each digit

if each comparison takes 40 milliseconds it should take you around 160 milliseconds.

So hypothesized stages are encode compare decide whether it is present or not and then respond.

Here you can see uh based on the number of items in the memory step one versus four the that is the exact amount of

time these people are taking. He created another experiment as well. In another experiments Turnber what he

did was he degraded the digits so that they became slightly harder to recognize. Okay. Everything is remains

the same as the experiment that we have just seen. He reasoned that if stimulus degradation affects the encoding state.

See, if you cannot see the stimulus, swell, it'll take you more time to encode. If it takes you more time to

encode, uh this time sort of adds in the overall time that you have to take across the four stages. Encode, compare,

decide, respond. So if stimulus degradation did affect the encoding stage, then the reaction

time would be longer than that on intact stimula. But the slope representing the relationship will not change. it'll

still be parallel to that same time. Just the stimulus degra stimulus encoding time has increased the

comparison etc does not change. But if stimulus degradation affects both encoding and comparison then the slope

would also change. All right. So the way the responses are say for example it was taking 38 millconds with every extra

digit it'll probably now take 40 more 40 millconds or 45 or 50 milliseconds in addition for comparison as well. Okay.

So if stimulus degradation would affect both encoding and comparison then the slope would also be slightly steeper. It

will take more time. Thus by adding an additional factor of stimulus quality he was able to determine whether encoding

and comparison were two independent stages or they sort of competed with each other or uh interacted with each

other. The results of this experiment actually showed that there was little difference

in the slopes. So it basically means there was little difference in the slopes between the degraded and the

intact stimula. Thus suggesting that stimulus degrading only affected the initial encoding stage but did not

affect the uh comparison stages and did not affect basically the final response. So that basically tells you that just by

adding one factor in the stimulus in terms of stimulus quality he was able to see at what stage of processing does

this particular factor affect. Now obviously Stunberg's research is also not without problems. It has its

own critiques and so on. But it demonstrated that reaction time is still a useful and a valid means for studying

the cognitive processes and therefore it plays a significant role in reviving the use of reaction time research in

cognitive uh science or in cognitive experimental psychology. The impact of uh Sternberg's reaction time research

can be seen in a very interesting comment made by Lacman and colleagues who says uh over the past decade it has

sometimes seemed as though half of cognitive psychologists in North America were devoting themselves to testing

Sternberg's theories testing how mental computations work how much time they take and so on. So in that sense you can

see more than half of of the uh psychologist in North America devoted themselves to conducting reaction time

studies and even now if you see uh large amount of work in experimental and cognitive psychology basically is uh

contingent on measurement of reaction times measure and computation addition subtraction of stages and so on.

I'll stop here. I'll continue this story in the next lecture. Thank you.