I. Star Reaction Time Test -- arrow and box variations
background:
This experiment demonstrates that attention can be focused on a particular location in space, and that attention can be moved independently of eye movements. The effects of attention are measured by showing that subjects respond to stimuli at attended locations more quickly than unattended locations; in other words, attention enhances the speed at which stimuli are processed. Furthermore, attention can be automatically drawn to a location in space. In this experiment, the outline of a box on either the left or right side of the screen is temporarily darkened (the cue in the box variation), or an arrow points towards either the left or right box (the cue in the arrow varation). In theory, the cue should draw attention to the location of the cued box, even though the subject's eyes remained fixed at the center of the screen. At a variable time after the darkening (or arrow display) occurs, an asterisk occurs in the center of either the left or right box (the target). If attention is drawn by the cue, and if attention enhances the detection of the target, then the following should be true: when the cue is valid -- i.e., the target occurs in the center of the box whose outline was darkened (arrow pointed towards) -- the subject should respond more rapidly than when the cue is invalid -- i.e., the target occurs in the box opposite to the cued box. The arrow variation is used to demonstrate that the cue does not have to occur at the same location as the target in order to see an attentional effect.
The effects of attention are dependent on the delay between the presented cue and target: as the delay increases, the reaction time benefits of the cue decrease. Inside the folder titled "further studies" is another version of the program titled "attention II". This program allows the user to specify 3 different delay values to be used between the cue and target presentation. Suggested delays are 100 ms, 450 ms, and 850 ms. The output computes the mean reaction time values for each different delay. To increase reliability, more trials are used for this program, which means it about 10 minutes to complete. The difference between the values is likely to be small, so in order to demonstrate the effect of delay more clearly, results from the testing of more than one subject can be added together. This averaging should offer the opportunity to talk about individual variability, population means, and statistical analysis.
II. Word Decision Task
background:
Words are not stored equally in the brain. In particular, words that are commonly used (high-frequency words) are more readily accessed than words that are less commonly used (low-frequency words). This frequency effect can be demonstrated by asking subjects to make a word/non-word decision about presented letter strings. Most subjects will make the decision more rapidly for high-frequency words than they will for low-frequency words. The frequency effect can also be demonstrated in a variety of other ways -- for example, subjects begin reading aloud high-frequency words more rapidly than low-frequency words, and they more accurately identify briefly flashed high-frequency words than low-frequency words.
A variety of explanations have been suggested to account for these word frequency effects. One of the more widely accepted explanations argues that words are represented by a pattern of activity across a set of neurons, and word recognition can occur only after a certain threshold of activation has been reached. In this model, a high-frequency word is recognized more rapidly because the neurons involved in representing the word are more strongly and reliably connected; thus, the threshold for recognition is reached more quickly.
Recent evidence from positron emission tomography (PET) studies of visual word processing suggest that the neurons involved in representing a visual image of a word are located in a region near primary visual cortex. (Primary visual cortex is located at the back of the brain, and it is the first cortical brain region to receive incoming visual information.) Patient studies suggest that if this potential visual word region is damaged (due to a stroke, tumor, etc.) then the ability to read will be lost, or severely impaired, though the ability to produce and comprehend spoken language can remain largely unimpaired. It is data of this sort which provides strong evidence that functions are localized in the brain -- i.e., each part of the brain is specialized for a particular type of operation, such as representing visual words.
further investigations:
Have your students write down their reaction time for real words, and the number of errors they made. Plotting each student's error and reaction time data on a graph of errors (accuracy) vs. reaction time (speed) should reveal a speed-accuracy trade-off effect: as reaction times decrease, the number of errors should increase. A sample graph is shown below:
In my experience, students tend to concentrate on how fast they went, assuming
that this means they are "better" at the task. Plotting the speed/accuracy
function can be a nice demonstration that speed is achieved at a cost.
(However, if you have students with very different reading abilities in your
class, you may not find such a speed/accuracy effect.)
I have enclosed a stimulus file with a longer list of words titled "word_stim-long" in the folder titled "further studies". In order to achieve a greater number of errors, I would suggest using this file by moving it from the "further studies" folder to the "programs" folder. The program will look for the "word_stim-long" file first, and if it doesn't find it, it will use a shorter file which is always in the "programs" folder. After you are done with the "word_stim-long" file, just move it back into the "further studies" folder.
You can also investigate priming effects. Priming refers to the influence of prior exposure to a word on subsequent performance. As mentioned in the background section, one model of word representation postulates that words are represented by a pattern of activity, and a certain level of activation is necessary to recognize the word. It is known that recent exposure to a word can enhance the speed at which a word is recognized. According to the model, this is explained by assuming that exposure to a word temporarily raises the baseline level of activation of the neurons involved in representing the word. If the word is presented again while the baseline activation is still enhanced, the threshold for word recognition will be reached more quickly. This effect is called repetition priming, since residual activation left from prior exposure to the word "primes" the neurons for another presentation of the word. Other types of priming effects, which can similarly be explained by activation changes, have also been found. For instance, prior exposure to a strongly associated word enhances the speed at which a related word can be recognized (e.g., prior exposure to the word "cat" enhances the speed at which the word "dog" can be recognized); this is called semantic priming.
Inside the "further studies" folder is another version of the word decision program, titled "word priming". In the first testing phase, students are asked to silently read aloud the displayed words. In the second phase, students are asked to make a word/nonword decision on presented words. For the word/nonword task there will be 40 words, divided by the program into two sets of 20 words (sets A and B), and 20 fake words. There are two different priming lists (list A and list B) used for the silent reading task. Both lists contain a set of 20 words which the students will not see in the word/nonword task. The remaining 20 words will be either the set A or the set B words used for the word/nonword decision task. Thus, students will be primed in the word/nonword phase for either list A or list B words, depending upon which list they saw during the silent reading task. In order to balance the priming study, you should tell half of the students to use list A when prompted by the program, and half to use list B. The program will keep track of which list was used, and as a final output will list the reaction time for the word/nonword task for primed ("old") words and for unprimed ("new") words separately. Since it is impossible to make set A and set B words equally difficult, an individual's reaction time for primed words may not necessarily be lower than his or her reaction time for unprimed words. In order to accurately determine how much priming lowers the reaction time, you will need to average the results from subjects who used list A with an equal number who used list B for the silent reading task.
This is another demonstration of how the brain is specialized (through experience) to process words (and sets of letters that look like words -- fake words). As explained above, there is evidence that a particular brain region may be specialized to represent printed words, allowing printed words (and fake words) to be decoded as a unit very quickly. For this reason, if you quickly flash a word (or fake word) all of the component letters can be identified, because the word as a whole is identified. However, for a set of quickly flashed letters that do not form a word or fake word, there is no such "word advantage" -- each individual letter must be decoded. If the letters are displayed very quickly, there may not be time to identify all of them. This "decoding" advantage present when a letter is presented in the context of a word can be demonstrated by forcing subjects to guess which of two letters was presented in a briefly flashed set of letters. Subjects will usually be most accurate when the letter was a part of a word, less accurate when the letter was a part of a fake word, and least accurate when the letter was part of a nonpronounceable set of letters. The difference in accuracy between real and fake words is thought to be explained by the same mechanisms which account for the frequency effects discussed above.
It is now widely agreed upon by scientists that at least two different types of memory exist. One type of memory, which has been called declarative or explicit memory, is the type that we are most familiar with: this is the memory used for remembering what we did yesterday, the names of our friends, where we parked our car, etc.. Another type of memory, which has been called procedural or implicit memory, is characterized by a benefit from previous experience which affects performance independently of conscious recall or recognition of that experience. Different brain structures are involved in each type of memory. Damage to the hippocampus, among other areas, causes deficits in declarative memory abilities, but leaves procedural memory intact. The brain structures involved in procedural memory are less clearly understood, but recent evidence suggests that damage to the cerebellum affects some procedural memory abilities, while leaving declarative memory abilities unaffected.
The jumping star program is designed to provide an example of procedural memory. Evidence that the task uses procedural memory comes from studies of patients with profound declarative memory impairments (due to stroke, alcoholism, etc.); these patients show normal improvement on the jumping star task, despite an inability to remember details about the task.
The program is divided into four task blocks. In the first and fourth
blocks, an asterisk jumps randomly to different locations, and the subject must
press a key which corresponds to the location as quickly as possible. Any
general improvement in the ability to do this task (e.g., improvement in
coordinating a finger movement in response to asterisk location) is measured by
a decrease in reaction time between the first and fourth blocks. In the second
and third blocks, the asterisk moves in a repeating sequence. Usually,
subjects are not aware that the sequence is repeating. Regardless of whether
the subjects are aware of the sequence, they are usually faster to press the
keys in the third block than they are in the first and second blocks. This
improvement is generally larger than the improvement between blocks 1 and 4,
suggesting that something else besides just a general improvement is affecting
speed. That "something else" is an implicit knowledge of the repeating
sequence -- i.e., even though they may not consciously realize there is a
pattern, their fingers, in a sense, do realize there is a pattern. As
experience with the pattern increases (i.e., block 3 versus blocks 1 and 2),
subjects respond more quickly. (See the figure below for an example of how
reaction times should change across blocks.
further investigations:
A longer version of this program, titled "jumping star II", is included in the "further studies" folder. For this version, a longer sequence is used, so that subjects are less likely to detect the pattern. In addition, subjects go through a set of seven blocks, rather than four. Since the pattern is longer, the learning of the sequence is slower, and you should be able to demonstrate a gradual improvement in reaction time across the blocks. Additionally, the investigator who initially developed this task (Dr. Mary Jo Nissen at the University of Minnesota) has demonstrated that normal learning of the task occurs only when subjects are concentrating solely on the task; if their attention is divided by forcing them to do two tasks at once, they do not learn the task at the same rate. You can demonstrate this with your students. Have roughly half of them listen to a Walkman (playing fairly loudly) when they perform the task. Ask those students wearing Walkman's to count how many times they hear the word "you" played on the radio while they are performing the jumping star task. Typically, the following results will be found:
1) Across all blocks, students who listened to the Walkman will have higher reaction times than those students who did not listen to the Walkman. This is because, in general, subjects go more slowly when they have to do two tasks simultaneously.
2) The reaction time of students who listened to the Walkman may decrease substantially across blocks 1-6. (In fact, their reaction time decrease, in milliseconds, may actually be greater than the decrease for students without Walkman's.) However, most of the decrease shown be students listening to Walkman's should reflect an improved ability to perform the two tasks simultaneously; they should not show much improvement due to a learning of the pattern sequence. Thus, whereas the reaction time between blocks 6 and 7 should go up for those students without Walkman's, the reaction time between blocks 6 and 7 should be roughly equivalent for students with Walkman's.
The performance differences between students with and without Walkman's can be made even clearer by expressing all reaction time values as a percentage of the reaction time in block 1:
%value of block X = (reaction time for block #X / reaction time for block #1) * 100.
V. Search Task
background:
There appear to be two very different ways by which a person can search for an object. In some cases, usually those for which a target object can be defined by a feature which is different from all other objects present (distracter objects), all objects can be assessed at the same time, and as a result, the target object can be located very rapidly. This type of search is called a parallel search. In such cases, the target is said to "pops out" from all of the other objects. As an example, imagine you were looking for a friend in a crowded room, and you knew that she was wearing a red hat. If no one else was wearing red, you would be able to find your friend very rapidly. In the other type of search, termed serial search, objects must be assessed individually, or in small subsets. This type of search usually occurs when a target object is defined by some combination of features which distinguishes it from distracter objects. As an example, imagine you were still looking for a friend with a red hat. If some of the others in the room were wearing red shirts, you would now have to search not only for the color, but also determine whether the red item was a shirt or a hat.
The search task program is designed to demonstrate the difference between parallel and serial searches by measuring reaction time. The basic principle is that the time taken to find a target will increase with the number of distracter objects when the search is serial, but not when it is parallel. Thus, subjects must look for a `T' mixed in with a varying number of `L's (a serial search), and for a `X' mixed in with a varying number of `L's ( a parallel search).
VI. Suggestions for Further Reading
Treisman, A.. Features and Objects in Visual Processing. Scientific American, November 1986, Volume 255, pp. 114B-125.
Mishkin, M. and T. Appenzeller. The Anatomy of Memory. Scientific American, June 1987, Volume 256, pp. 80-89.
Squire, L. R. Mechanisms of Memory. Science, 20 June 1986, Volume 232, pp. 1612-1619.
Petersen, S. E. P., P. T. Fox, A. Z. Snyder, and M. E. Raichle. Activation of Extrastriate and Frontal Cortical Areas by Visual Words and Word-Like Stimuli. Processing of Visual Words. Science, 31 August 1990, Volume 249, pp. 1041-1044.
Posner, M. I.. Localization of Cognitive Operations in the Human Brain. Science, 17 June 1988, Volume 240, pp. 1627-1631.