[From Bruce Abbott (2014.01.13.0945 EST)]
Rick Marken (2014.01.12.2130) –
Bruce Abbott (2014.01.12.1930 EST)
BA: But I disagree with your assertion that “conventional experimental psychology is built on an illusion,” because there are plenty of experiments in psychology that do not manipulate disturbances to control systems as the independent variable and observe the resulting behavior as the dependent variable.
RM: Could you name a couple?
BA: How about psychophysics? I manipulate the brightness of
a light and ask you to indicate when you first perceive it.
RM: The independent variable is the brightness (it’s manipulated independent of the subject), The behavior that is the dependent variable is the indication (yes or no) regarding whether it is perceived. The subject is asked to control for a relationship between their yes/no indication and the perception of the light. The independent variable is, thus, a disturbance to a controlled variable (if it weren’t the subject would probably say nothing on each trial). The dependent variable corrects for the effect of that disturbance. So this is not an example of an experiment where the experimenter “does not manipulate disturbances to control systems as the independent variable and observe the resulting behavior as the dependent variable”.
If PCT is correct, then any experiment in which we ask the participant to do something while we manipulate an independent variable (such as the brightness of a light) will involve a control system. The question is not whether there is a control system at work, but whether the experiment’s data tell us only about the environment while the experimenter believes that it is providing information about the participant – the “behavioral illusion.”
So, let’s see if that’s true in the case of determining the “dark adaptation curve.”
To make the experiment easier to understand in control terms, we’ll give the participant a knob to turn. By turning the knob, the participant varies the intensity of a light-source. The light is initially off and the participant is told to turn the knob until she can just barely perceive that the light is on. She is to do so immediately after hearing a brief tone.
Now the experiment begins. The experimenter turns off the room lights, plunging the experimental cubical into total darkness. The tone sounds immediately, and the participant quickly adjusts the knob until she just barely perceives the light. The experimenter records the intensity of that light, then turns the room light back on, rotates the knob back to the “off” position, and waits for the participant to readjust to the light. This procedure is then repeated, this time after having the participant wait for one minute in darkness before the experimenter sounds the tone. This process continues until the threshold for detecting the light has been recorded for time-in-darkness ranging from 0 minutes to 50 minutes. Here’s the resulting curve:
The question is, does this curve tell us something about the participant’s visual system, or is it an example of the “behavioral illusion,” telling us only about a feature of the environment?
According to you, Rick, it just shows us the inverse of the environmental function – roughly speaking, the inverse of relation between the participant’s setting of the knob and the intensity of the light that the participant is trying to detect. But that is a smooth curve, with no inflections in it. The inverse of that function does not have an inflection point in it. The curve of our data points does. It can’t be the inverse of the knob-light intensity function.
So let’s analyze the experiment in terms of control theory. We’ve instructed our participant to adjust the knob until she just barely perceives the light. In PCT terms, we’re asked her to set a reference for light intensity at the threshold value for perceiving the light. If she follows our instructions, then she will turn the knob until the perceived intensity of the light closely matches that threshold value. By having her repeat this adjustment after varying delays since the room lights were doused, we determine how that threshold value changes over time in darkness. The resulting data, plotted as above, show that this threshold value decreases (the participant is becoming more and more sensitive to light as time elapses). In addition, that decrease in threshold initially follows an approximately exponential function out to about 10 minutes, where it almost levels off at a still rather high threshold value. At that point something unexpected happens: the curve starts to fall again, and over the next 40 minutes falls (again following an approximately exponential function) until it levels off at a very low threshold value.
These changes in threshold over time have been traced to changes in the sensitivity to light of two varieties of photoreceptors in the participant’s retina – cones and rods. The cone cells provide the basis for color vision but do not increase their sensitivity to light much in darkness. (This is the reason why colors fade away as darkness descends.) Immediately after lights-out, the cones are more sensitive to light than the rods, so their sensitivity determines the first part of the curve. The rod cells do not provide color discrimination but their sensitivity increases greatly in darkness. (They provide us with night vision.) At about the 10-minute mark their sensitivity has increased enough to match that of the cones at that same point in time; after that the rods are more sensitive and determine the remainder of the curve.
Does the experiment engage the participant’s control systems? Yes. Do the data tell us nothing about the participant’s visual system, but only about the relationship between the knob setting and the intensity of the light? Hardly.
BA: Or trans-magnetic stimulation studies, demonstrating, for example,
that applying a strong magnetic field near the margins of the occipital
cortex can cause the colors perceived in one half of the visual field to
disappear during the stimulation (which disables the functions being carried
out in that part of the cortex).
RM: This is similar to the threshold experiment. The independent variable is the magnetic field; the behavior (reporting whether or not the color appears) is the dependent variable. The subject is asked to control for correctly saying what color (or no color) is seen on each trial. So the magnetic field is a disturbance to that variable – affecting the color (or lack thereof) that is seen – and the subject’s response compensates for that disturbance (by saying color or no color when asked). So again this is not an example of an experiment where the experimenter “does not manipulate disturbances to control systems as the independent variable and observe the resulting behavior as the dependent variable”.
You have this way of changing the ground of the argument without really saying so – a bit of the magician’s misdirection of attention. Again, the point is not to find a conventional psychological experiment that does not involve a control system, but rather to provide examples of conventional psychological experiments in which the data tell us something useful about the participant rather about the environmental feedback function.
Trans-magnetic stimulation temporarily “paralyzes” the cortical neural networks in the vicinity of a rapidly alternating magnetic field. If you place the magnet near the rear of the head (where the occipital lobe of the cortex is found), at just the right location, you can disrupt the processing being carried on by the neurons there. The participant sees the color disappear from one half of the visual field of both eyes. If the right cortical hemisphere is being subjected to the trans-magnetic stimulation, the color will disappear from the left half of the visual field of both eyes. This demonstrates two things: (1) the right hemisphere processes the data from the left half of the retina of both eyes, and (2) the area stimulated has something very important to do with the processing of color in those halves of the visual scene. When that processing is disrupted, our conscious perception of color is lost for that portion of the visual field.
Again, we discover something important about the participant from this study. The data do not reflect the “behavioral illusion.”
BA: Or studies demonstrating how it is possible to “implant” a false memory.
RM: You would have to explain how the study was done. I can’t tell from this description.
BA: It doesn’t matter, the same kind of analysis applies as in the previous examples.
BA: Or studies indicating the heuristic
“shortcuts” people use when dealing with situations where carrying out the
required operations for an accurate analysis would be too difficult or
time-consuming.
RM: Again, you would have to explain how these studies were done. But assuming that the independent variable in these studies was the difficulty of the problem to be analyzed (solved?) and the DV some measure of the heuristics (how many, perhaps) that were used to solve them, then again you have an experiment where the experimenter manipulates disturbances (difficulty level) to a controlled variable (state of the problem with a reference of “solved”) and observes the resulting behavior (heuristics) as the dependent variable.
BA: I could go on . . .
RM: Please do. But I’ll give you a hint: Virtually any experiment where subjects are given instructions, if they are human or subjected to “establishing operations”, like being starved, if they are non-human, is one where the experimenter manipulates disturbances to a controlled variable as the independent variable and observes the resulting behavior as the dependent variable. It’s all purposeful behavior. I once saw a book on operant conditioning of a dead person; I think if you can find that book you will find a lot of examples of experiments where the experimenter “does not manipulate disturbances to controlled variables as the independent variable and observe the resulting behavior as the dependent variable”. Because dead people don’t control.
Again, the argument was not whether participants in conventional psychology experiments control (they do!), but whether the data from such experiments necessarily fool the experimenter into thinking she is learning something about the participant, when in fact she is only discovering the inverse of the environmental feedback function and is thus falling prey to the “behavioral illusion.” No doubt there are some experiments in which that happened, but I’ve provided a small sample of typical experiments in which the data do inform us about the participant and not the feedback function.
Bruce
