INSTRUCTIONS FOR THE FIRST LAB REPORT ASSIGNMENT
To ease you into the task of writing lab reports, I've made this first assignment a partial report: the Introduction, Method and Reference sections of the report on Experiment I: Reaction Time.
The Introduction and Reference sections should be straightforward since they don't involve the actual procedure or results. Make sure that you append a photocopy of the first page of the journal article that you refer to in your Introduction.
The Method section requires some creativity since you have not yet chosen your subjects or actually done the procedures.
You will have three subjects (plus or minus one) in your experiment.
You know that you will be one subject so you can provide the information asked for in the Lab Manual Subjects section.
Make up two more subjects; you can substitute the data for the real live subjects when you hand in the full report.The Apparatus section should look pretty much like the Lab Manual; there's no need to quote it since there's only one way to report that information.
The Procedure section is another exercise in creativity.
The Lab Manual describes how the experiments should be conducted.
Based on this information, report what should have been required of each of your three subjects as if they had actually completed the experiment.
Be sure that you give a complete account of what happened to the subjects and what they were required to do.
Do not report what you did to the apparatus, or what the subjects' results were.
Paul K. Brandon
Psychology 207, sec. 8
August 26, 2004
Your report will begin with
an Abstract. As I've already described it, an Abstract is a brief
summary of an article, which can be read without having the
article itself.
Your Abstract should tell what the general topic of interest is, who or what the experimental subjects were, and a brief description of your experimental procedures (a sentence or two should tell the reader what the basic structure of your experiment was). Make sure that it is clear how your experimental conditions differ from the baselines.
The Abstract should also summarize your results, and state what you concluded from the results of your research. You should include both the specific results of your experiment and some more general conclusions about basic principles of behavior.
The Abstract should fit on the cover page of your report, which should look like this page. Again, the best way to learn how to write an Abstract is to read them in real journals!
(Your Introduction must
start on a new page)
Reaction time, the time between the presentation of a stimulus and
the response to it, was one of the first aspects of human behavior
to be measured. It is said that early astronomers noticed a
consistent difference of a fraction of a second in their reported
star sighting times. They came to the realization that people
could not respond instantly no matter how well trained and
prepared they were. As a matter of fact, a second is quite a long
time, behaviorally speaking (how far does a car travel in one
second?).
People often use the word reflex when they talk about reaction time. People who respond quickly to a stimulus are described in everyday language as having 'quick reflexes'
However, in behavioral terms a reflex is based on an unlearned or innate (genetically based) stimulus/response relationship &endash; one that would be shown by a person completely new to a situation. This is referred to in the text as respondent or elicited behavior. That is, a reflex (whether unconditioned or conditioned) is controlled only by its antecedents -- it is not affected by its consequences (review pp 10ff in the Chance text for a discussion).
Pressing a key would be an
unconditional reflex only if we could get it to occur in
any human being, regardless of their language or
experience, simply by presenting the appropriate stimulus.
A conditional (or conditioned) reflex would start with an
unconditioned reflex and transfer the control to a new stimulus
(called a conditional stimulus) by pairing that stimulus
with the original unconditional stimulus described
above.
What we are studying here
is really a well-learned operant under stimulus control; something
that is quite different! Chance (1998) points out that reaction
time is an example of latency; one of our basic measures of
behavior. You might take a look at what he says about these
topics.
In your introduction, please include at least one reference from a
professional journal (the library does have a few -;) on the
general topic of reaction time. Try to find one that has some
relationship to this experiment and/or to your results. Do
find one of your own; not one that someone else in the
class has used. Summarize, in your own words, the
findings of this study in a sentence or two.
The experimental question being
addressed concerns the effects of various arrangements of stimulus
presentation on the time that it takes a human subject to respond
to those stimuli.
This will be accomplished by the following experiments
a) reaction time to a simple visual stimulus, b) reaction time to a simple auditory stimulus, c) reaction time in a situation involving two stimuli and two responses, and d) reaction time involving two textual stimuli and two responses.
NOTE: I've been using 1.5 spacing until now, and am shifting to single spacing to make this manual easier to read.
Your reports should be double-spaced &endash; no more than 4 lines per inch &endash; to allow me space to make comments on your papers.
Subjects
The subjects will be members of
the Intro Behavior Analysis class at Minnesota State University.
We can assume that all class members are human beings, but
we do need to know their ages, sex, (a simple yes or
no is not sufficient!), handedness, and experience
(describe it briefly) with this sort of task in the past. Without
this background information, we cannot generalize our results to
the behavior of other human beings in other situations.
Identify all subjects by their initials (e.g.: "subject JQS
was 97 years old, female and left handed. She had no previous
experience with this type of experimental procedure").
You are a subject in this experiment, and must be reported
as such!
Apparatus
A Macintosh iMac microcomputer with an LCD display and a pair of
headphones. The timing program written in HyperTalk
Bring in a couple of sheets of
carbon paper to make duplicate copies of your raw data!
(Don't put this statement
in your Apparatus section, though, it is not something that
affects your subjects' behavior).
*** NOTE *** I have put this in future tense; you will use past tense, as you are reporting an experiment which has already taken place.
Also -- you are usually expected to use your own words when writing reports. In some cases, however, it is very difficult not to use the same words as the lab manual. The Apparatus section is one of those cases.
Procedure
Report the exact procedure that you actually used for each subject -- including yourself. Do not 'clean up' your procedure in the report; tell it like it was!
Please note that this is a "cookbook" for you to use in running your experiment. Therefore, I am presenting it in outline form, and often use conversational first-person language. You should use proper English prose. Strunk and White (2000) is a very good resource for the clear and concise use of language!
Do not report machine settings or what the experimenter does on the keyboard; other experimenters may have different equipment. What counts is what happened to the subject!
In order to be able to make a statement about whether a given experimental manipulation has produced a given effect, we must make sure that nothing else changes in the same way as our experimental variable. Therefore, since we want to draw conclusions about comparisons between reaction times to light and to sound stimuli, we must make sure that no other differences are correlated with the changes in stimuli.
Two obvious possible confounding
variables are handedness and practice. To control for these, you
must have some subjects use their dominant (usually right) hand
for light, and other subjects use that hand for sound. Similarly,
some subjects must be tested for their reaction time to sound
first and then light, while others have their reaction time tested
to light first and then to sound. One possible experimental
design (arrangement of experimental conditions) is illustrated
below. An easy way to organize your experiment is
to fill
in your subjects' initials in the "SUBJECT" column of the table
below:
|
SUBJECT |
1st SERIES OF TRIALS |
2nd SERIES OF TRIALS |
|
1: |
Right hand / Light |
Left hand / Sound |
|
2: |
Right hand / Sound |
Left hand / Light |
|
3: |
Left hand / Light |
Right hand / Sound |
|
4: |
Left hand / Light |
Right hand / Sound |
NOTE: On your data sheet, "order run" should tell you whether that set of trials was the first or second set run for that particular subject.In this design, the dominant hand is sometimes used to respond to light and sometimes to sound. In the same manner, reaction time to light is tested first for some subjects, while reaction time to sound is tested first for others. Therefore, if there is a consistent difference between reaction time to sound and to light, it cannot be due to either handedness or to practice, since neither of these variables were consistently paired with a particular stimulus.
For each series of trials, keep a record of:
- who was the subject,
who was the experimenter,
what the stimulus-response (S-R) pairing was, and
the order in which the two stimuli were presented to that subject.
In the example shown below, JQS was the subject, ABC was the experimenter, JQS responded to a light stimulus with his right hand (the stimulus-response pairing), and was tested for reaction time to light first and for reaction time to sound second (the order of stimulus presentation).
For each trial record: which stimulus was presented, which response was made (these will be the same for each of the fifteen trials in a series for the first two experiments), and what the reaction time was.
Set up your data sheets before you start the experiment. Your data sheet for
Experiment A: Reaction Time to a Visual Stimulus Presented Alone
might look like the following:
Subject: JQS |
Subject: ABC |
Subject: XYZ |
Note that the individual columns for stimuli and responses are not really necessary here, since they don't change during each set of 25 trials. However, in the following two experiments you will need these columns!
You should each have a copy of the raw data for all subjects (including yourself) run by your lab team. Since hand-copying introduces a source of possible errors, you should either photocopy the data sheets, or make carbon copies as you record. The latter is recommended; it's cheaper!
Also note:
- The purpose of recording data
is so that you have a record of what occurred during each trial
of the experiment.
- You should have the
information necessary to make a statement like
- "on trial number 24
the subject responded to the sound stimulus with her left
hand with a latency of .58 seconds:."
- "on trial number 24
the subject responded to the sound stimulus with her left
hand with a latency of .58 seconds:."
We're still working with the Academic Computing Center (the new proprietor of our computer lab) to get the new machines set up.
Clicking on an icon of a folder labeled either Psychology or IntroBA in the Dock (the strip of icons at the bottom of the screen) will open up a folder with the class programs in it. - You should have the
information necessary to make a statement like
If this does not happen, then you might have to go into the Applications folder.
Either way, double-click on the Exp1--Reaction Time icon to run the program for this experiment. I will demonstrate in class.
You may want to adjust the volume of the sound in the headphones (first making sure that they're plugged in ;-) by using the speaker volume keys in the upper right hand corner of the keyboard (next to the CD eject key).
1. Follow the computer's prompts to specify the stimulus and response to be used.2. Instruct the subject to place the index finger of the specified hand so that it is just touching the specified response key: <Q> for the left hand and <P> for the right hand.
3. Ask the subject if he or she is ready.
4. Click the mouse in the Do it! box. The computer will wait for a random time interval between one and five seconds, to prevent the subject from timing the experimenter's behavior and anticipating the presentation of the stimulus.
5. The computer will now present the stimulus, and measure to the nearest hundredth of a second the time that it took the subject to respond correctly to it.
6. Record the reaction time in hundredths of a second (e.g., .17sec).
7. Repeat steps 3 - 6 until you have reached a total of 25 trials for each S/R pair: 25 trials each for Experiments A and B, and 50 trials for Experiments C and D.
Note: the statement ".17sec" is read: "seventeen hundredths of a second"! The statement ".17 hundredths of a second" means "seventeen ten thousandths of a second" ! Bear this in mind when you are labeling graphs and tables! Most conventionally, your unit of measurement will be seconds, stated as decimal fractions to two decimal places (if you don't know what this means; ask!!).
If you are averaging data
(calculating the mean), round your results to this
level of precision. Do not imply that you are measuring
time to the nearest thousandth of a second, just because your
calculator gives you seven digits to the right of the decimal
point! Do not imply more precision of measurement than you
actually used!
One convention for deciding whether to round a "5" up or down is
to round to the nearest even value. Thus, .175sec would be rounded
to .18sec; .165sec would be rounded to .16sec. This results in
your rounding up half of the time and down half of the time.
Therefore, no systematic bias is introduced into your data through
the rounding process.
- Again a reminder:
- I am writing
a Lab Manual instructing you how to do an experimental
procedure.
- You should write a Lab Report which describes what happened to a subject, not what you did to experimental apparatus.
- I am writing
a Lab Manual instructing you how to do an experimental
procedure.
Experiment A: Simple reaction time to a visual stimulus
In this experiment you will
measure each other's reaction time to a light stimulus: a picture
of a light bulb presented on the computer screen.
Follow the general procedure outlined above, remembering that some
subjects must use their dominant hand for light and others for
sound, and that some subjects must have their reaction time to
sound measured first and reaction time to light measured
second.
Therefore, reaction time to a visual stimulus must not be the first experiment actually run for all of the members of your lab team.Your Procedure section must make clear the order in which each subject did the light and sound reaction time experiments.
Experiment B: Simple reaction time to an auditory stimulus
This experiment measures reaction time to a sound stimulus; a beep. For each member of the lab team, your response should be made with the hand that you did not use to respond to the light. Again, note that some of you will run this experiment first, and then test reaction time to a visual stimulus.
Experiment C: Two-stimulus / two-response reaction time
This experiment is a combination of the two preceding experiments. We will test the reaction time for each subject to both light and sound in a situation where the subject does not know which stimulus will be presented on a given trial, and thus which response will be required. He or she must use the stimulus to decide with which hand to respond.
Enter the stimulus/response
pairings for the subject. Each subject should use the same
S/R pairings as in the first two experiments. The computer will
set up a sequence of 50 trials, twenty five (25) each to light and
sound in a random order. It is best to use a different sequence
for each set of trials; the computer will do this.
Lay out your data sheets as before. Note that now you will
have to note different stimuli and responses for each
trial.
Follow the same general procedure as in the simple reaction time experiments. Now, both the stimulus and response may change after each trial.
Record reaction times for
correct responses only!! If a subject does manage to
make the wrong response, hit him/her over the head with the
nearest blunt instrument :-). The computer will not record
that trial.
One reason for using the median (the middle value)
rather than the mean (the average) as a summary
statistic is that the median is not affected by individual
extreme scores caused by mistakes.
Experiment D: Textual stimulus / two-response reaction time.
The procedure for this experiment
is the same as for the two-stimulus/two-response experiments
except that each subject will now be instructed to respond to a
word ("LEFT" or "RIGHT") presented in the middle of
the computer's screen. This will allow for comparisons between
reaction time to simple stimuli and reaction times to more complex
stimuli involving written language.
NOTE: when reporting your Procedure, you need not go through all
the stages of explanation that I have in the lab manual. Start by
completely describing the procedure for the first part of the
experiment. Then, describe changes made to this basic
procedure.
For the choice and textual reaction time experiments, separate each set of 50 reaction times into two sets of 25.
Draw three (3) graphs for each subject including yourself run by your lab team -- one for reaction time to light, one for reaction time to sound, and one for reaction time to textual stimuli -- as illustrated in Figure 1.
The graphs should have captions that enable the reader to understand each graph without having to check back into the body of your Results section. Each line on the graph should have a descriptive label; not just a letter or number. The axes of the graphs should be labeled and have units indicated. See Figure 1 below for reaction times to auditory, visual and textual stimuli.
Most of you will draw your graphs electronically. If you do it by hand, please use graph paper and a ruler to plot your graphs. It is permissible to use different colors to differentiate different lines on the graphs. In published journals this is done by using different line and point patterns.
After you have plotted the graphs of the individual reaction time trials (plotted in the order in which the trials were run, of course), calculate the median reaction time for each subject under each condition. To do this, rank order each set of 25 reaction times. Note that this means that you must have separated the 50 trials in the two response experiments into two sets of 25 trials each; one to each stimulus.
The median reaction time is the middle one, or the thirteenth longest (or shortest) out of 25. Report these median reaction times in the body of your Results section, using a tabular format as in Table 1.
If you have access to a computer with a spreadsheet and drawing program you may use that to do the graphs. The one limitation is that the graphs must meet the requirements of scientific data presentation, such as having all axes begin at the zero point. The example in this manual was done using the AppleWorks spreadsheet and drawing modules on a Macintosh (see the Appendix of this experiment for detailed instructions). This program is available on the computers in our lab, and you are welcome to use it when available. Once you've entered the data into a spreadsheet, the program will draw the graphs. You can also use the sorting function of the spreadsheet to find the medians!
Pay attention to the proportion of the axes. They should resemble the examples in the manual. The data should not be so compressed or expanded that trends are obscured.
One common problem with some computer graphing programs is that they insist on putting zero values at the beginning or end of a series.
Since you will not have any real zero reaction times (it's not physically possible ;-), white these out!.
Table 1
Median reaction times (in seconds) to visual, auditory and textual
stimuli for all subjects.
SIMPLE CHOICE TEXTUAL SUBJECT LIGHT SOUND LIGHT SOUND LEFT RIGHT ABC . 20 .19 .35 .32 .34 .35 CDB .18 .21 .32 .32 .30 .25 XYZ .22 .17 .25 .27 .26 .25
- NOTE: This figure was done using the drawing module of AppleWorks. It could also be done from the spreadsheet, but you'll have less control over things like patterns and spacing.
After you have calculated the median reaction times, average these medians by conditions (find the average of the median reaction times for all of your subjects under each condition). This will give a summary of the overall relationship between reaction time and your experimental conditions. Present these mean (average) reaction times in a histogram, as shown in Figure 4.
In the text of your Results section, as well as in the Discussion section, refer the reader to your Figures when you comment on them.
For example:
"See Figure 1 for the individual trial data on reaction time for subject JQS."
The Discussion section is commentary on your results.I've tried to integrate the Discussion sections in this Lab Manual with Chance's text.
Your Discussion section must contain answers to all of the questions below:
1. What is the clearest, most
obvious result of your experiment in terms of the effects of
your experimental manipulations on your experimental
variables?
Differences between individuals is not an
experimental variable in this study.
2. What might be the implications of your findings for human behavior in the 'real world'? (If you're not sure what I mean by "implications", look the word up in a dictionary. This is course true for anything that I say in the Manual).
3.Were your results unexpected (again, emphasize the effects of experimental variables, not individual differences)? If so, what might be the reason?
- If your procedures were different from those specified in the lab manual, how might these changes have affected your results?
4. One way in which science extends the generality of its findings through the process called systematic replication. This is the repetition of an experimental procedure with a change made in one variable only. What other experiments might answer questions raised but not answered by these experiments?
5. What does it mean to "measure" something? Look up the word measurement.
- Some people might say that we
have measured the speed of human thought; how long it
takes a person to think. Can we really draw this conclusion
from our data?
- What did we actually
observe?
See what Chance says about thought and thinking (for instance, pp 32 and 37).
- What did we actually
observe?
6. Why don't we say that we are measuring the speed of the subjects' reflexes?
Again, review the introduction in the Manual, as well as what Chance has to say.
7. Are the mean reaction times obtained by averaging the data from several subjects representative of the results from individual subjects? Comment on this on the basis of your data.
- Do the patterns of individual
medians in Table1 correspond to the
means of the median reaction times plotted in
Figure 4?
- Are there any trends in the individual data presented in Figures 1 - 3? See Chance, page 77 for the type of analysis that I am referring to.
- One particular trend that you might look for is a practice effect; an improvement in performance over the course of a single session of 25 or 50 trials.
A basic point about
science:
It is similarities among subjects that are most important,
not differences. Look for the effects of the experimental
variables that are common to all subjects, rather
than for differences among your subjects.
If you had any computer problems, tell me which computer it
was.
- I'd also welcome any comments
about the design and presentation of this experiment.
-
Strunk, William Jr. and White, E.
B. (2000). The Elements of Style (4th edition).
New York: MacMillan Publishing Co.
I have listed this since I referred to it in the body of my paper! Any other published work that you mentioned in your report -- most usually in either the Introduction or the Discussion -- should be listed in alphabetical order using the standard format shown in this Manual.
In the body of your report, cite sources that you referred to by the author and year of publication only, rather than by numbered footnotes. The author's name and year of publication often provide enough information for you to recognize a reference in a field that is familiar to you.
To append something means to add it onto something else.
This section is for anything that didn't fit into the main body of your report. In this course, the Appendix will contain your raw data recorded directly from your observations during the experiment.
Check your data sheets to make sure that you have provided all of the information specified in the results section !!!
Your Appendix must also include a photocopy of the first page of your reference, and (if you printed out an electronic version of an article), a page with a figure or table on it.
to produce a graph like Figure 1 in the manual.
You may use other programs such as Excel if you like, but you must produce a proper scientific graph that looks like the one shown in Figure 1.
|
Use the
MavDISK server to save your files while you
are working on them. Go to <http://mavweb.mnsu.edu/>
to activate your MSU account if you have not already done
so.
|
|
|
Note the difference between saving a file on the computer that you are working on and your MavDISK account, which lives on a server in another building. To save a file onto more than one MavDISK account, you must first save it into the SHARED folder in the USERS folder. Once the
file is moved into this folder it is accessible to anyone
working on that particular computer. The other alternative is to save it directly into the public folder or your MavDISK account where other people can get it. However, you can't easily transfer a file from one MavDISK account to another. I'd strongly recommend saving the file locally -- on the Desktop of the computer you're working on -- first; then and only then saving it into a MavDIsk account. Finally, the best way to save your files might be to purchase a USB Jumpdrive. These are inexpensive (they start at about $20 at the MSU Microcomputer Store), small,(they fit on a keychain) quick and easy to use. |
|
1. Launch AppleWorks from the dock and OPEN a spreadsheet file.
2. Enter the data for one subject in the format illustrated above.
- The first row provides the labels for the legend; the first, fourth and seventh columns provide numbers for the X (horizontal) axis.
- Highlight the columns that you wish to graph, including the rows and columns with axis labels..
3.Under the "Options" pulldown menu at the top of the screen choose <make a chart>.
|
|
4.Under "Gallery",
|
|
|
we'll do that later in the drawing module. |
|
|
6.Under "Series":
|
|
|
7.Under "Labels":
|
|
|
column". This will make the column headings 9. Click on "OK". This will draw your graph.
double-clicking on the graph. |
10. Select and COPY the graph.
11. OPEN a drawing file. Select
"Page view" in the "View" menu.
12. PASTE the graph into it.
13. Repeat for the other two graphs for that
subject.
14. Resize the graphs by dragging on the lower
right corner while holding down the <shift> key so
that all three graphs fit on the page.
15. Type in your Title and Caption by clicking on
the "A" symbol. If you don't see it, click "show tools"
in the "View" menu. Drag the Title to its location at the
top of the page.
16. Type in your axis labels the same way. Rotate
the vertical axis label by using the "Rotate"
command under the arrange menu. If you can't find
it, make sure that the 'arrow' symbol is selected at the
top of the tool menu.
17. Remove the borders around the graphs by
setting the line width under the pen symbol in the
tools menu to "none".
PRINT THE GRAPH!!
You can also use the
AppleWorks program to draw the bar graph that summarizes the
experiment.
I used the drawing module to draw rectangles of the appropriate
heights -- you could also enter the data in a spreadsheet and use
the graphing program, selecting the "Bar" option.
Graphing can also be done in other programs such as Excel (Micro$oft Office), but you may have to work harder to get a plain black-on-white graph that looks like this one. Office is designed for business graphics, where the point is not to produce an unbiased presentation of data!
ABSTRACT (write your own)
This experiment will deal with
human behavior in situations where more than one response is
possible and different responses have different outcomes. Your
Introductoin should begin with a statement like this.
Choice behavior may be influenced by factors other than the
correctness of the choice. In this experiment we will examine two
possible sources of control over choice behavior: 1) Extrasensory
Perception (ESP), and 2) The consequences (events that follow it)
of behavior.
As with the previous experiment, please provide at least one reference from a professional journal (not the popular press, magazines, books or web sites) that is related to at least one of the above topics.
The choice situation that we will use is based on the same apparatus that we used in the previous experiment.
The procedure for testing ESP will be in the form of a 'guessing game'; the subjects will be required to guess which of the two responses is correct on any given trial, with no way (hopefully) of knowing which of the two responses is in fact scheduled on a given trial. Our data, therefore, will consist of records of which response (left or right key) was programmed on a given trial, which response was actually made on a given trial, and whether or not the subject was correct on that trial.
To test for the effects of consequences on choice behavior, we will use a primitive video game that again requires the subject to choose one of two responses.
While behavior analysts do not usually use inferential statistics, sometimes they are helpful. In this case we are dealing with the variability of behavior. Chance discusses this in his unit on interpreting data. This can be helpful; we take advantage of the variability of behavior to shape new behaviors. However, it can make it difficult to tell the difference between random variability and a real effect. Ultimately, this question should be answered by improving the quality of our experimental control so that the effect of our experimental manipulation is clearer, but if that is not possible we may have to resort to a statistical analysis.
Note that this may leave us with an effect so small that, while it is real, it is not of any practical value. Sometimes we talk about the difference between statistically and clinically significant differences.
Also note that we are using a single subject ABA experimental design (as described by Chance), not a group design. Rather than collecting data from a large number of subjects, we are collecting a large quantity of data from a single subject. We will repeat this process for more subjects to establish the generality of this result. If you are interested in some typical group data, some results for this class are available online at http://www.mnsu.edu/dept/psych/Psych207f2kESP.html.
The number of highlighted false positives (apparently significant differences expected from random variation) are about what would be expected in a sample of this size.
Subjects
For the first set of experiments, your subjects will be the members of your lab team including yourselves, as in the first project. As in the previous experiment, include yourself as one of your subjects. Provide all of the necessary information, including experimental histories and knowledge of the purpose of this experiment!
For the second (differential consequences) set of experiments, each lab team will be responsible for bringing in two experimentally naive subjects; subjects who do not know the purpose of the experiment. It's a good idea to schedule these subjects so that the first one comes in about fifteen minutes after the start of the period (to allow for questions), and the second about forty five minutes later.
Apparatus
A Macintosh iMac microcomputer, programmed in HyperTalk and a set of headphones.
Procedure
Again, check the instructions in the program.
The first set of experiments will take the form a guessing game, with the subjects trying to guess which of two responses will turn off a tone generated by the computer
As the computer will have to know which experimental condition (experimental or control) is to be in effect so that it can set up conditions accordingly, you must start by entering this information. The computer will prompt you!
- Again, these instructions are a cookbook.
- You should report what the computer does to the subject;
not what you do to the computer.
The computer will use a different random order for each set of trials, to control for possible learning effects (actually, it cycles through several different random sequences).
Start with the
experimenter facing the computer,
and
the subject oriented so that they can not see the screen
(swivel it <the computer screen, not the subject>
around).
For the basic procedure, the subjects will place their index fingers on the <Q> and <P> keys. The experimenter will tell him or her that a trial is about to begin, and the computer will sound a tone. After the tone stops the subject should press one of the two keys. If the correct key was chosen, you will hear a 'running chord' of three tones. If the wrong choice was made, the computer will emit a rapid series of beeps. After each trial, the computer will inform the experimenter which response was programmed and which response was actually chosen by the subject. The experimenter will then record the responses programmed and made.
It is recommended that you run each subject under all three conditions (Baseline, ESP, Baseline) before you run the next subject. This is the most efficient way to complete the experiment, and it minimizes the effects of extraneous variables.
Experiment A: Baseline
This will be a control condition, run as described above. There will be a total of 50 trials; 25 each of left and right responses programmed in a random order. There should be no way in which the subject can predict which of the two responses will be programmed as correct, so their performance should be at about the 'chance' (guessing) level of .50 (50% correct).
A reminder: the subjects should not be able to see the computer screen.
Experiment B: ESP
Now, we will introduce the experimental variable of extrasensory perception. The experiment will proceed as before except that now the computer will show the experimenter which response will be correct on each trial, and the experimenter will concentrate on the response which the subject is to make. If the subject has ESP and can read the experimenter's mind, he or she should be able to do better than the chance level of .50 expected from simple guessing.
Experiment C: Return to baseline
This is an additional control session. In particular, it is a control for practice effects. If a subject shows an improvement between the first baseline and the experimental sessions, it could be due either to the effects of the experimental variable (ESP), or simply to practice. It the latter is the case, the improvement should be maintained through the second baseline (return to baseline) session. On the other hand, if the improvement is due to the experimental variable, performance should show a decrease during the return to baseline, when the experimental variable of ESP is removed.
Experiment D: Baseline
To examine the effects of consequences on choice behavior, we will use the medium of a video game. To be specific, the subject will be instructed that he or she can earn points on a video game by pressing two keys on the keyboard.
The subject will be confronted with a computer screen on which is drawn a five-by-five grid, with an image of a rat in the upper left corner. The subject is to move the rat to the lower right-hand corner by pressing the <Q> and <P> keys. A point is earned each time the rat reaches that corner. Thus, we have two levels of outcome; the immediate outcome of moving the rat one position, and the outcome of gaining points.
Technically, both of these outcomes are conditioned reinforcers. That is, they are effective because they have been paired with other reinforcing outcomes.
This procedure also provides an example of a reinforcement schedule (look it up in the text if you're curious. Every eighth key press moves the rat (a fixed ratio eight schedule), and every eighth rat image move gains a point (also an FR 8 schedule of reinforcement).
The subjects will be told that you are measuring how long it takes them to earn five points. In fact, we are mainly interested in which key they respond on. The computer will be programmed in this first condition to alternately move the rat across and down when a total of eight key-pushes have been accumulated, regardless of which keys have been pushed. Thus, there is no reason to use one key or another. This is a control or baseline condition measuring the subjects' tendency to choose between their left and right hands.
In order to minimize random responding, we will add a second contingency (behavioral requirement): the computer will not respond to responses separated by less than about one-half second.
When a subject has reached five points, the computer will end the game.
Record the time that the subject required to get five points. The computer will also display the number of responses that the subject made on the left and right keys (indicated by 'X' and 'Y' so as not to bias the subjects). Record these also, making sure that you record ALL the information necessary to answer the questions who did what, when.
The procedure for displaying these data are illustrated below:
Experiment E: Differential consequences
Now we will introduce the experimental variable of differential (selective) consequences for right-hand key responses to see if this has an effect on the subjects' choice behavior.
To do this, the computer will be programmed to required twice as many right-hand responses as left-hand ones to move the rat.
ill only count half as much as left-hand responses, and thus be less likely to move the rat and gain points, fame and reinforcement. Technically, we have doubled the size of the fixed ratio reinforcement schedule for the right hand only, to see if this increase in response cost (hint) has an effect on behavior.
Of course, you will not tell the subject this! You will see if they find it out from experience. Proceed exactly as in the baseline condition.
Experiment F: Return to baseline
This control session has a similar purpose to the one in the ESP experiment. However, in addition to controlling for practice effects, we can also look for delayed effects of consequences, even after the differential consequence has been withdrawn and both keys are equally effective.
When you have completed the second set of experiments, 'debrief' your subjects by explaining to them the purpose of the experiment, and answering any questions that they might have. This is a necessary part of any experiment that involves human subjects!!! It is also good experimental procedure as it provides an interview setting in which you may get additional information from the subjects on topics not specifically provided for in your procedure, such as their guesses as to just what the heck was going on!
In particular, you might ask what strategies they used, and what they thought the 'rules of the game' were. You'll find this information helpful when you write your Discussion section.
For each of the two sets of experiments, your Results should tell the reader:
1. What proportions were compared,
2. What statistical test was used to make the comparison, and
3. Which comparisons yielded statistically significant differences,
Your Results should take the form of proportions; decimal fractions ranging from zero (.00) to one (1.00). In the ESP experiments you are interested in the proportion of correct responses, while in the differential consequences experiments you concerned with the proportion of responses made with the right hand. Therefore, for the ESP experiments, divide the number of correct responses by the total number of responses, which should equal 50.
In the differential consequences experiments, divide the number of responses made on the right-hand key by the total number of key responses (Y ÷ (X + Y)).
Present this data in the form of
Tables, as illustrated in Table 1 and Table 2 below.
This displays the data in a form where the results for each
individual subject can be easily seen, and comparisons can be
readily made.
Round the proportions to a fixed two decimal places,
as illustrated below:
with and without the use of ESP.
|
Subject |
Baseline |
ESP |
Baseline |
|
ABC |
.53 |
.68 |
.57 |
|
XYZ |
.55 |
.70 |
.57 |
|
RIP |
.43 |
.41 |
.51 |
Report what statistical test was used!
For example: "Possible significant differences between observed and predicted proportions were analyzed using the Test for the Significance of the Difference Between Proportions" (more on this later...).
In the first set of experiments, for each subject, compare the proportion of correct choices under a given set of conditions with the proportion correct (.50) that would be expected from chance or guessing
"subject XYZ did significantly better than would be expected from random guessing when using ESP".
As this example indicates,
the only comparisons which are directly reported are those which
are statistically significant.
Your Appendix should contain your work sheets, which should
show all of the z values that you calculated, as well as
the specific comparisons between proportions that produced these
z scores. Put these comparisons and z scores in a
Table in your Appendix.
>>>> DO YOUR OWN WORK! I realize that many of you will calculate significance as a team, but I expect that each of you will use your own work sheet (not photocopy someone else's) and include it in your Appendix. You are personally responsible for everything that you hand in under your name!
For the second set of experiments (differential consequences), report the proportions of right-hand responses (the number of responses on the <P> key divided by the total number of responses) in a similar table, as below:
|
Subject |
Baseline |
Differential |
Baseline |
|
CDB |
.60 |
.45 |
.50 |
|
QRS |
.75 |
.68 |
.62 |
|
FFV |
.40 |
.45 |
.27 |
Here, for each subject, you will compare pairs of observed proportions; you are looking for the effects of an experimental manipulation on the proportion of right hand responses
- (e.g.; "Subject FFV made a significantly smaller proportion of right-hand responses during the second baseline condition than during the first").
As before, report all significant differences in the text of your Results section; do not report minor variations that might reasonably be expected due to random sampling error.
This, of course, raises the question of how we know how much difference we need to call an effect "significant". For this task we use statistics! What we need is a "probability statement"; what are the odds that a given difference is "real" and not simply the result of random variance.
What we want to be able to say is that the odds are at least 100 to 1 against a given effect being due to chance. If random variation produced an effect of that size no more than one time out of twenty, we can say with reasonable confidence that something happened!
In this experiment we will use the Test for the Significance of the Difference Between Two Proportions, as described in Appendix A. It is fairly easy to calculate with a common electronic calculator, but to make it even easier, I have a program on the Mac that will do the dirty work for you.
It is a basic rule of science that you must assume that something does not exist until you have been shown otherwise. The burden of proof is on the person who is making the claim that something exists.
If an experiment does not produce any statistically significant results when you apply a significance test, you must report that there was no effect! There is no such thing as "almost significant." You can not say things like "while not significant, the difference between X and Y....".
If there is no statistically significant difference, you must treat the apparent differences as meaningless variations in the data.
One can never prove a negative outcome (no difference between two values), but unless we have good evidence, we must assume that there is no difference; that a variable has no effect on behavior.
1. For both the ESP and the Consequences experiments you must state whether your evidence supports the existence of the phenomena that you are studying. Was there any evidence of ESP?
- Of the effects of consequences on behavior? Review the discussion above on what a real effect is.
2. Again, did anything interesting happen? What were the effects of the E S P experimental condition (the experimenter looking at the programmed response) on the subjects' performance?
- Remember, when we do a statistical test we are finding out what the odds are that we will see a given effect just from chance alone. These odds are greater than zero, so we would expect some false positives: apparently significant differences that are due only to random variation. This should happen about one time out of one hundred at the level of statistical significance that we are using.
3. If you had found any significant effects, what factors other than E S P might have produced your results (in other words, alternative explanations)?
- Even if you did not find significance, speculate on what might cause apparently significant evidence of ESP.
4. In the differential consequences experiments, what was the effect of the increased relative work requirement for the right-hand responses? You might talk about functional relationships here.
- Note that there are two places at which an effect might be seen: during the differential consequences phase, and after the differential consequence has been administered!
- If there was no
effect, why might this be?
- (Hint: reinforcer effectiveness).
5. What is our technical term for a consequence of behavior which results in a decrease in the frequency of occurrence of a specified behavior?
- Do you have any evidence in your data that this occurred? Describe it!
If you found a statistically significant effect in the opposite direction what might be the cause?
6. Did your subjects
report any awareness of the purpose of the experiment
while it was going on?
- During the debriefing after the experiment?
- Did their behavior change without their being aware of the change?
- Could they describe the
variables controlling their behavior?
- Do you think that people must be aware of (be able to describe) what controls their behavior for behavior to change?
-
