Abstract
The use of the Propeller Experiment Controller to create inexpensive teaching laboratories in behavior is described. Pre-written programs are provided for unique classroom experiments in habituation and classical conditioning. Several programs are also provided to recreate the traditional student operant conditioning laboratory. Video tutorials are also available to guide the user to effectively use the Propeller Experiment Controller for behavioral experiments.
This article offers readers interested in the experimental analysis of behavior an inexpensive, versatile, and powerful experimental controller suitable for teaching laboratories. Our design is based on the Parallax Propeller microcontroller (Parallax Inc.; Rocklin, California). In contrast to experiment controllers available from companies such as Lafayette Instruments and Med Associates that cost thousands of dollars and are cumbersome, the Propeller costs fewer than one hundred dollars, is easily portable, powerful, and readily adaptable to a wide range of situations. A laboratory can literally be placed in the palm of your hand and carried from office to classroom in your pocket. See Fig. 1 for an image of the Propeller Experiment Controller.
Two experiment controllers in one hand. Top: Propeller Platform DNA. Bottom: Parallax QuickStart.
For over two decades our laboratory has been fighting increasing use of computer simulations such as Cyber Rat ((AI)2; Winter Park, Florida) and Sniffy the Rat (Graham, Alloway, & Krames, 1994) as a replacement for hands-on, inquiry-based learning experiences in behavior (Abramson, 1986, 1990; Abramson, Curb, Barber, & Sokolowski, 2011). Such products do not allow students to experience the nuances associated with the study of behavior and reduces the student to playing what amounts to a video game. In a study specifically designed to compare the experiences of working with live animals and computer stimulations, students found the live animal demonstration to be more realistic and more valuable as a learning experience than a computer simulation of classical conditioning (Abramson, Onstott, Edwards, & Bowe, 1996). The authors find it interesting that no one seriously questions the silliness of using a video game to train clinical psychologists, yet video games are routinely employed to replace animals in the teaching process.
We also believe that the lack of meaningful student experience with animals contributes to the decline of comparative psychology as a viable psychological enterprise. Cyber Rat and Sniffy the Rat are not suitable for a comparative psychology course because they only focus on emulating the behavior of a single species. What is needed to generate students' interest in comparative psychology are exercises that are truly comparative and experiential (Abramson, Hilker, Becker, Barber, & Miskovsky, 2011).
In this paper, we provide a brief introduction to programming and connecting the Propeller to various response devices such as a levers and photocells. We also describe how the Propeller can be used to activate lights and reinforcement dispensers. Our software that enables the Propeller to be used as an experiment controller is included, as well as pre-written programs for experiments in habituation, classical conditioning, and operant conditioning. The software package can be found by clicking here (http://www.amsciepub.com/doi/suppl/10.2466/07.08.IT.2.2). We discuss the use of these programs in a classroom setting. Video tutorials are also available to guide the user to effectively use the Propeller for behavior experiments. These tutorials can be found by clicking here (http://www.amsciepub.com/doi/suppl/10.2466/07.08.IT.2.2). To fully appreciate this paper some basic electrical skills are necessary. Experience with a breadboard, soldering iron, and voltmeter will be beneficial. Programming experience is not required to use the experiment programs we provide. However, some basic programming skills will allow the user to create novel experiments that go far beyond those described in this paper.
Overview of the Parallax Propeller
The Parallax Propeller microcontroller has 32 input/output (I/O) pins. Each I/O pin can be connected to some other device, either an input or an output. Input devices are typically used to record the activity of a subject and commands from the user and may include levers, photocells, and user interfaces. These devices are not limited to a simple digital on-off input; the I/O pins are also able to interface with a wide variety of devices that enhance the Propeller's functions. In addition to detecting simple events such as the number of times a lever was pressed, we have also connected the Propeller to other inexpensive devices to record temperature in an apparatus, use echolocation to detect the location of a subject in a runway, and to detect the faint glow from bioluminescent organisms.
Output devices are typically used to operate stimuli such as lights and food hoppers. The Propeller can also create more complex forms of output than the simple turning on and off of a device. The Propeller is powerful enough to generate audio and video signals. Audio generation is useful in any experiments related to auditory discrimination or conditioning. No separate tone generators are required; simply connect the Propeller to a speaker. Generating video is useful for tasks using visual stimuli without requiring the user to install an array of lights. ATV monitor or LCD screen can also be connected to the Propeller to allow the user to see data in real time. User interfaces can also be created through a combination of devices or through touch screens to provide control over the variables in an experiment without additional programming. These types of interfaces are great for student devices and may be used to allow the student to select specific values of experimental variables such as the inter-trial interval, and reinforcement schedule.
The Propeller is also very easy to use. We have experience with several programming languages, almost always related to experiments or data analysis, and have found the Propeller's language, “Spin”, to be one of the easiest. Another feature that makes the Propeller easy to use is its unique multi-core design. Each of the Propeller's 8 processors, called cogs, can complete tasks independently of the others. This makes programming complex tasks easy, and even makes it possible for the Propeller to run several unrelated, independent experiments at the same time.
Complex tasks are made simple through the use of prewritten “objects.” The Propeller is very well supported by a variety of users in hobby, industry, and robotics. Objects to deal with a variety of tasks are written and often made freely available. The objects function to perform a variety of tasks from writing to SD cards, to using infrared sensors to detect distance to an object, to generating video and detecting input on a touch screen. There is an object for everything. This allows users to write programs faster as they do not need write every function themselves. It also allows users to perform complicated functions that may be beyond their current level of programming skill.
The Propeller as an Experiment Controller
The previously mentioned characteristics make the Propeller an excellent experiment controller. We have used the Propeller in our laboratory to conduct investigations in the learning of bees, bioluminescent algae, pigeons, planarians, and rattlesnakes. To make experiments easy, we developed a set of objects dedicated to conducting experiments and collecting data. These objects provide a framework for future experiments. Our experience with the Propeller in the laboratory suggests that it would be an excellent device for classroom use. To further encourage classroom use we also developed 20 programs for classroom experiments.
Using the Propeller for Classroom Experiments
In this section, we describe turn-key programs for several laboratories in habituation, classical, and operant conditioning. These laboratories are suitable for a variety of courses including those in introductory psychology, psychology of learning, and comparative psychology. Moreover, the laboratories can be used by students in conducting their own independent investigations as part of, for example, a directed reading course.
The habituation and classical conditioning laboratories are unique. There are currently no automated classroom demonstrations for these two classes of conditioning. Our habituation laboratory allows the student to vary such training variables as the number of habituation trials and time between trials. The classical conditioning laboratory allows the student to explore not only the effect of training variables but also various arrangements of conditioned and unconditioned stimuli.
We have also included a series of operant conditioning exercises. These exercises repeat the classic demonstrations so familiar to a previous generation of students. In this laboratory, students learned to shape their animal's behavior and conduct simple demonstrations in acquisition, extinction, discrimination, chaining, and schedule effects.
For those readers unfamiliar with using microcontrollers or connecting input and output devices we suggest that you read the sections on preparing the Propeller for experiments and attaching devices in Appendix A. We also created several video tutorials that may be helpful.
Habituation
In this section, we will describe how to use the Propeller to create a student demonstration in one of the more simple types of learning, habituation. Habituation refers to the reduction of a response to some stimulus as that stimulus is repeatedly presented. Although habituation is a simple form of learning compared to classical and operant conditioning, it shares many characteristics with these higher-order forms of learning including discrimination, generalization, and spontaneous recovery. Therefore, habituation experiments are an easy and practical way to demonstrate complex learning characteristics. In the following paragraphs, we will describe how to present stimuli, select training variables such as stimulus duration and inter-trial interval, and record responses using habituation programs we created. The instructions in this section are more detailed than instructions related to classical and operant conditioning experiments and serve to familiarize the reader with the Propeller and the programming paradigm we use throughout all the provided experiments.
First, open the habituation program in Propeller Tool or BST. You will notice the large document is broken into color-coded blocks of text. Much of this text is code to run a variety of habituation experiments. Additional text found within brackets or following quotation marks contains descriptions of how the program works. The reader is free to read the code along with the comments to gain a better understanding of how to create other programs for classroom experiments. The general function of the program is to repeatedly present a stimulus, while recording responses from the subject. The program also allows the user to provide specific values of training variables.
Before conducting habituation experiments, a little information must be provided to the program. Focus on the CON block starting at line 21. This section contains constants that are used by the program to detect responses, activate stimuli, and implement specific training procedures. First, you will need to provide some information about the SD card that the Propeller Experiment Controller uses to save data. See Appendix A for instructions on installing an SD card. Provide the pin numbers of the connections to the SD card on lines 33–36. Each connection is named following SD card conventions. If you use the Propeller Platform DNA, or a QuickStart with a Human Interface Board, the pin connections will be as follows: DO = 0, CLK = 1, DI = 2, CS = 3.
Next, provide the pin numbers connected to the apparatus on lines 50–54. The response pin refers to the pin connected to the device that detects a subject's response. The stimulus pin refers to the pin connected to the stimulus device, such as a vibrating motor or a light. The house light pin is optional. It refers to a light traditionally built into an apparatus that turns on when the session starts and turns off when the session ends. The diagnostic LED pin refers to an optional connection to a diagnostic LED. In all our programs, this LED will start flashing after the experiment is complete to signal that it is safe to remove the SD card. Although this device is optional, we highly suggest using it so that the SD card is not removed prematurely. Finally, the escape pin refers to an optional connection to a button that allows the user to end the experiment if the subject does not habituate. If one of the optional devices is not desired, set the pin number to a pin that is not connected to another device.
By default, the program will name the stimulus presentations “Stimulus” and any activation of the response device “Response” in the data file. The user is free to change these names in the “SetVariables” method on lines 156 and 157. Simply change the word “Response” or “Stimulus” into whatever description is desired. Note that the quotation marks are important. Only change the text within the quotation marks.
Training variables for the habituation program can be set in the CON block on lines 41–44. As learning occurs as a function of experience, the subject will need multiple habituation trials. Too few trials will not allow for a demonstration of habituation. We suggest using at least 15 trials. It is possible that some individuals may not habituate within this number of trials. For this reason, we included an optional mastery criterion. The mastery criterion is the number of trials without any response that must occur before the experiment ends. This allows an experiment to continue until a subject demonstrates habituation. For example, if you set the mastery criterion to 5, then 5 consecutive trials must occur without the subject responding to the stimulus before the experiment ends. However, if you set the mastery criterion to 0, the experiment will end as soon as the initial trials are complete, without regard to the subject's responses. See Place and Abramson (2008) for an example of using a mastery criterion in habituation of the rattle response in rattlesnakes.
Set the stimulus duration on line 43. Note that the duration is listed in milliseconds, as this is the base unit of time used by all our programs. Multiply minutes by 60,000 to quickly translate minutes into milliseconds. Longer stimulus durations may produce faster habituation than shorter stimulus durations. If the stimulus is long enough, you may also observe that the subject begins responding at different times during the stimulus. Initially, the subject may respond as soon as the stimulus starts. As trials pass, it may begin to respond closer to the middle or the end of the stimulus before habituating. A quick analysis of the data.csv file may make these changes easier to notice than by direct observation alone.
Another important training variable is the time between stimulus presentations or inter-trial interval. Inter-trial interval can be set on line 44. Generally, the shorter the inter-trial interval, the faster habituation occurs (Thompson & Spencer, 1966). One interesting classroom experiment is to use short inter-trial intervals with one group of subjects, and long inter-trial intervals with another group of subjects. This type of simple demonstration allows students to prove for themselves that training variables, in this case inter-trial interval, greatly affect learning.
Variable stimulus durations and inter-trial intervals may produce different rates of habituation than fixed stimulus durations and inter-trial intervals. Variability is likely to cause slower habituation. However, you may notice that habituation generalizes better to novel stimulus durations and inter-trial intervals after receiving training using variable durations. To explore these factors, we created a second habituation program that will allow the stimulus duration and the inter-trial interval to be randomly selected from a user-specified range. Open the variable habituation program. Note that it is very similar to the habitation program, except that you will need to provide the minimum and maximum duration of the stimulus presentations and inter-trial interval. If the minimum and maximum duration are identical, no variability will occur. Also note that although the general format is the same, some line numbers of constants changed slightly. From the user's perspective, the programs are otherwise identical.
Intensity of a stimulus is also an important factor in habituation. Animals habituate to weaker stimuli faster than stronger stimuli (Thompson & Spencer, 1966). Additionally, an individual that has habituated to a weak stimulus, such as a low volume noise, may still react to a stronger stimulus, such as a loud noise. A third habituation program titled “Variable Intensity Habituation” allows the user to specify a variable level of stimulus intensity in addition to variable levels of stimulus duration and inter-trial interval. As with other variable factors, the variable levels of stimulus intensity are generated randomly from user provided minimums and maximums. Additionally, an optional second stimulus intensity level can be used. On lines 54 and 55, you will notice familiar constants regarding the number of trials and mastery criterion for habituation. These refer only to the first stimulus intensity level. After the number of trials has passed and any mastery criterion is met, the program will begin presenting the stimulus at a second intensity level. On lines 57 and 58, you can set the number of trials and mastery criterion for this second intensity level, or set both to 0 if you do not wish to use a second intensity level. Both levels of stimulus intensity are set on lines 65–68. The second level of intensity enables experiments such as habituating a response to a low intensity, then increasing the intensity to observe if the response remains habituated. For example, if an earthworm habituates to a weak vibration from a motor, will a strong vibration elicit a response?
One important note about this program is that it will not work with devices controlled by relays. The program uses a method called pulse-width modulation (PWM) to create varying levels of stimulus intensity. This involves rapidly turning a device on and off thousands of times per second, creating an illusion of changes in intensity such as dimmer light levels or slower motors. Relays cannot operate this quickly, so this program is only suitable for devices connected directly to the Propeller's I/O pins or to devices connected through transistors. For relay-controlled devices we recommend the use of potentiometers placed between the device and its power source. This will allow students to control the intensity of the device manually.
A variety of habituation experiments with multiple species can be conducted using these programs. Planarians are inexpensive to procure and maintain and can be used to demonstrate a variety of learning principles. A brief air puff from a relay-controlled aquarium pump can elicit body contractions and extensions. Air puffs and vibrating motors can also be used to elicit the audible hissing response of Madagascar hissing cockroaches, body contractions in earthworms (Abramson, 1990), eye stalk retractions in crabs (Abramson & Feinman, 1988), and head-withdrawal responses in turtles. Acoustic stimuli can produce startle responses in a variety of species but may not be suitable to classroom demonstrations if multiple experiments are being conducted in a small space. Although these responses are easily observable, some are difficult to detect electronically. We suggest requiring students to press a pushbutton each time the subject responds. This form of manual recording maintains students' attention while still providing more detailed and accurate data than paper-and-pencil reports.
Simple habituation experiments with students can also be conducted. For example, a strong vibration from a vibrating motor held in the hand may cause a noticeable reaction in a student. A student can press a pushbutton to record a response each time they are startled. Students can also work in pairs, with one student acting as the subject and another student acting as the researcher. Have the researcher student press a pushbutton when they observe a response. The students can then alternate. This exercise is also useful because the students must carefully consider how they operationally define a response to the vibration so that they can accurately record responses.
When the experiments are completed, the program will generate a spreadsheet file titled “data.csv” that can be opened and analyzed in programs such as Microsoft Excel. Graphs can also be constructed for visual analysis of the data. See Fig. 2 for an example of the data file. All of our programs create this kind of data file. It is important to note that any file named “data.csv” will be overwritten when new data is saved. Be sure to remove the SD card, save the data to the computer, and then change the file name to something more descriptive before starting a new experiment. Advanced users may take advantage of some methods provided in Experimental Functions to save spreadsheets with custom names.
An example data file automatically generated by the habituation program. The instance column refers to the number of the event, such response instance 1, the first response. The onset and offset columns refer to the start and stop times of an event, respectively. Duration refers to the duration of that instance of an event, while total duration refers to the combined duration of all instances of an event. Inter-event interval refers to the time between events of the same type. Finally, total occurrences refers to the total number of times an event occurred in a session. When the file is produced, it is initially sorted by event type. In most spreadsheet programs, the data can be sorted by row. In this example, the data has been sorted by onset, so that the events appear in chronological order. The event, instance, and onset rows indicate that fewer responses were elicited by the stimulus as the experiment progressed. Also note that the inter-event interval of the response, or the time between responses, increases from approximately five seconds to eleven, then twenty-two seconds.
A maximum of 1,000 instances of each type of event can be saved to the spreadsheet. If more than 1,000 instances of an event occurred during the session, an error message will be saved to the SD card. To create a spreadsheet, you can use the Python program included in the Propeller Experiment Controller package. See Appendix C for a brief description of the Propeller Python Interface program.
Classical Conditioning
Classical conditioning is a more complex form of learning in which an animal learns an association between a conditioned stimulus (CS) and an unconditioned stimulus (US). The CS is an originally neutral stimulus that comes to elicit a response only after it is repeatedly associated with a US that already produces that response. In this section, we will describe how to use the Propeller to create classroom demonstrations and experiments in classical conditioning. For instructions regarding setting up the Propeller and related software, please refer to the instructions in the habituation section and Appendix A.
Open the classical conditioning program in Propeller Tool or BST. As with other programs you will need to provide pin numbers for the SD card, stimuli, and response devices. The general function of the program is to present the CS and US during a series of trials. See Fig. 3 for an example data spread sheet created by the classical conditioning program. Set the trial length on line 50. On lines 51–54, you can set the time within the trial that the CS and US will start and stop. For example, if you set a trial length of 10 seconds, a CS that starts at 0 and ends at 10 will occur during the entire trial. A more likely interval would be for the CS to start at 0 and end at 5, while the US starts at 2 and ends at 5. This would cause the CS to start as soon as the trial begins, the US to start 2 seconds later, and both stimuli to end 5 seconds into the trial. After the final 5 seconds of the trial elapses a new trial will start. See Fig. 4 for an illustration of these constants.
An example data file automatically generated by the classical conditioning program. In this example, the data has been sorted by onset, so that the events appear in chronological order. Note that the response occurred after the US for the first 7 trials, and then began occurring after the CS but before the US on the remaining trials.
A diagram illustrating the constants in the classical conditioning program. The user provides a trial length, and the start and stop times of the CS and US. In the delay conditioning example, the CS_Start time occurs early within the trial and the US_Start occurs at a slight delay. Both stimuli end simultaneously as the CS_Stop and US_Stop times are identical. In the trace conditioning example, the CS_Start and CS_Stop times are the same as the previous example. Although the US has the same duration as the delay conditioning US, the US_Start and US_Stop times occur much later in the trial. The time between CS_Stop and US_Start can be considered the trace interval. Adjusting the start and stop times of both stimuli can create many other conditioning procedures.
As you can specify the start and stop times of both stimuli within a trial, a wide variety of conditioning procedures can be created. Many classical conditioning procedures are considered forward conditioning procedures in which the CS precedes the US. Forward condition procedures are generally effective because the CS predicts US. One type of forward conditioning is delay conditioning. In delay conditioning, the CS starts, then the US starts after a small delay, and then both stimuli end simultaneously. To create a delay conditioning procedure make sure that CS_Start is less than US_Start and that CS_Stop is equal to US_Stop. For example, with a trial length of 10 seconds, set the CS_Start to 0 seconds, the US_Start to 2 seconds, and both the CS_Stop and US_Stop to 4 seconds.
Another form of forward conditioning is trace conditioning. In trace conditioning the CS starts first, then the US starts after the CS ends. The time between the end of the CS and the start of the US is known as the trace interval. To create a trace conditioning procedure make sure that US_Start is greater than CS_Stop. The difference between these two constants is the trace interval. For example, with a trial length of 20 seconds, set the CS_Start to 0 seconds and the CS_Stop to 2 seconds. Then set the US_Start to 4 seconds and the US_Stop to 6 seconds. In this case, the trace interval is 2 seconds. Generally, the longer the trace interval, the more experience is required before the CS elicits a response. Other conditioning procedures such as simultaneous conditioning and backward conditioning can also be conducted by adjusting the start and stop times of the stimuli.
You will also notice several trial types listed on lines 56–58. Acquisition trials refer to the pairing of CS and US as in the previously mentioned procedures. During acquisition, the animal is acquiring or learning the association. The primary purpose of the classical conditioning program is to present the CS and US at the user-specified times during a user-specified number of acquisition trials. You can also use an optional number of inhibition trials, where the CS is presented by itself before the acquisition trials begin. As the CS is repeatedly demonstrated not to predict anything in inhibition trials, acquisition can take longer once the CS and US are paired. After the acquisition trials end, you can use an optional number of extinction trials where only the CS is presented. In extinction trials the association between the CS and US previously learned during acquisition becomes extinguished because the CS is no longer associated with the US.
Another method to study classical conditioning is discrimination training. An animal will learn to respond to a stimulus that is associated with the US, the CS+, and will learn not to respond to a stimulus that is not associated with the US, the CS-. A separate program titled “ClassicalConditioningDiscrimination” can conduct these types of experiments. The format of the program is similar to the previous classical conditioning program; you will need to set many of the same constants. This program is distinct in that there are no inhibition, acquisition, or extinction trials. Instead, the number of trials provided on line 58 reflects the number of CS+ and CS- trials. The total number of trials is double this number. At the start of each trial, the program randomly determines if the trial will be a CS+ trial, in which the CS+ and US are presented, or a CS- trial, in which just the CS- is presented. Alternatively, the program can use a pseudorandom selection so that the same trial type does not randomly occur many times consecutively. The number entered on line 59 for ConsecutiveLimit refers to the maximum number of trials of the same type that can occur simultaneously. Increase ConsecutiveLimit so that it exceeds the number of trials for completely random selection.
With these programs, many classical conditioning experiments can be conducted. Often an apparatus used to study habituation can also investigate classical conditioning. For example, the body-contraction response of the earthworm can be explored in either a habituation or a classical conditioning procedure. In the habituation procedure, vibration is repeatedly presented. In the classical conditioning procedure, some other stimulus, such as a light, acts as a CS that predicts the vibration (Abramson, 1990). The eye retraction of crabs in response to a puff of air can also be studied in either a habituation or a classical conditioning procedure. For crabs, light vibrations can act as a CS (Abramson & Feinman, 1988). For these two experiments, aquarium air pumps and vibrating motors can easily be controlled by relays. The responses, however, are more difficult to detect automatically and we again suggest that students activate a pushbutton to record responses.
Honeybees are also exceptionally good subjects for classical conditioning procedures as they learn relatively quickly and can be collected in large numbers. One common area of study in honeybee learning is the proboscis extension reflex. When a bee's antenna contacts sugar water, the proboscis automatically extends to drink the solution. This reflex can then be conditioned to occur to a wide variety of odor CSs. The procedure is simple enough to be conducted manually (Abramson, 1990) but can also be fully automated (Abramson & Boyd, 2011).
Classical conditioning experiments can also be conducted with students. For example, the same vibrating motor used in student habituation experiments can also be used as a US in classical conditioning experiments. An LED can then act as a CS. As before, students can work individually or in pairs and press a pushbutton to record responses. With the correct classical conditioning procedure, the students may begin to startle when the LED turns on before the motor activates.
Operant Conditioning
One of the most popular teaching demonstrations widely used by the previous generation of psychology students is the standard operant conditioning pigeon or rat laboratory. In this laboratory, students learned to shape an animal's behavior and conduct simple demonstrations in acquisition, extinction, discrimination, chaining, and schedule effects. In this section, we will describe how to use the Propeller to recreate traditional operant demonstrations and experiments. For instructions regarding setting up the Propeller and related software, please refer to the instructions in the habituation section and Appendix A.
The first demonstration in many operant conditioning laboratories is shaping. Often the target behavior shaped by the students can then be used as a foundation for further explorations in learning. Open the shaping program in Propeller Tool or BST. The general function of the program is to provide a reinforcer any time the subject activates the response device, or anytime the student activates another device to reinforce an approximation to the target behavior. Although both actual responses and approximations can produce reinforcement, the data spreadsheet makes a distinction to allow comparisons between the two types of responses and reinforcement. As with other programs you will first need to provide pin numbers for the SD card, stimuli, and response devices. The response pin refers to the pin connected to the response device activated by the subject, such as a lever for a rat. The shaping pin refers to the pin connected to a button or lever the student will activate to manually provide reinforcement for approximations.
Our operant experiments follow a free operant paradigm, where the subject is free to earn as many reinforcers as possible within a fixed session duration. Set the session length, in milliseconds, on line 42. When reinforcement is provided, our programs activate the reinforcement device for a fixed period of time. During this period of time, the subject can respond more, but cannot earn any other reinforcers. Set the reinforcement length on line 43. This duration can be adjusted to emulate traditional pigeon feeders, where food is presented for several seconds. Alternatively, you may need a brief reinforcement duration. Some reinforcement devices such as pellet dispensers can deliver a small amount of pellets when they receive only a few milliseconds of current. For this type of device, you can set the reinforcement length to be only a few milliseconds long.
The continuous reinforcement program functions identically to the shaping program except that there is no input to manually reinforce approximations. Once the student has established a response, you can switch from the shaping program to the continuous reinforcement program. After several sessions of continuous reinforcement, stable patterns in responding will likely emerge. Then you can switch to a program with a different contingency to demonstrate the effect of contingency on behavior.
The operant discrimination program provides reinforcement for a response only in the presence of an SD (discriminative stimulus), but never in the presence of an SΔ (S-delta). The program randomly alternates between SD and SΔ presentations. You can specify pins for the SD and SΔ devices on lines 62 and 63. If you want to use an absence of a stimulus as an SD or SΔ, provide the number of the pin that is not connected to another device. For example, if you want key pecks to be reinforced only in the presence of a red light (SD), and never to be reinforced when no lights are on (SΔ), connect the red light to the Propeller and provide the pin number to the program. Then provide an unused pin number for the SΔ. The SD and SΔ can have distinct durations. Set these values on lines 47–48. An optional time out interval can be set on line 49. During this period neither the SD nor the SΔ will be presented. As with the previously described classical conditioning discrimination, the SD and SΔ are presented randomly. However, you can set a consecutive limit to use a pseudorandom selection on line 52. An alternate version of the operant discrimination program that functions identically with slightly different code structure is also provided for readers interested in learning to create experiment programs.
We created a chaining program that will provide reinforcement only if a series of responses occurs in a specific order. For this program, a chain of either two or three responses can be used. On line 42, set UseThreeResponses to true to enable a three-response chain. Set it to false to use a two-response chain. On lines 47–49, you can set the pins for all three-response devices. The order the devices are listed here reflects the order the devices must be activated to produce reinforcement. Even if you set the program to the two-response chain mode, it will still monitor the third response device. Set the third device pin number to an unused pin if you do not have three response devices. From the user's perspective, the program is otherwise set up as any other operant program.
Exploring schedules of reinforcement is another common exercise in student operant laboratories. In the fixed ratio schedule (FR) of reinforcement programs the subject must respond a fixed number of times before receiving reinforcement. Set the FR requirement on line 42. A variety of exercises can be created just from changes in FR schedule. For example, transitioning from continuous reinforcement to a low FR schedule will result in an increase in response rate. However, if the FR schedule is initially too high, the animal may stop responding. FR schedules can also be gradually increased to very high levels. See Fig. 5 for an example data sheet created by the fixed ratio program.
An example data file automatically generated by the fixed ratio program. In this example, the data has been sorted by onset, so that the events appear in chronological order. Note that three responses must occur before reinforcement is produced, indicating a FR3 schedule. For traditional operant conditioning experiments, response rate can be derived by dividing the total occurrences of the response by the session length.
We also created a variable schedule (VR) of reinforcement program. The VR program is identical to the FR program except that the response requirement is chosen randomly after each response from a user-provided range. On lines 43 and 44, you can set the minimum response requirement, VRmin, and the maximum response requirement, VRmax. All other factors can be set as they were in the FR program. Typically VR schedules produce a higher rate of response with fewer pauses than FR schedules with similar response requirements.
Fixed interval (FI) schedules of reinforcement can be used to explore an animal's ability to sense time. In this program, reinforcement is only provided when a response occurs after a fixed time interval passes. Responses before this interval elapses have no effect. Set the FI interval on line 41. Typically, as an animal learns the time interval, responses will occur more at the end of the interval than the beginning or middle. This trend can be revealed by creating scatter plots of response onset on the x, versus response instance on the y. This type of plot is very similar to B. F. Skinner's cumulative record.
A variable interval (VI) program is also available. Like the VR program the VI program functions nearly identically to its fixed counter part. The user only needs to add a minimum and maximum interval on lines 41 and 42.
Fixed time (FT) and variable time (VT) also deal with time intervals. However, they are distinct from FI and VI schedules in that the subject is not required to make a response. In the FT schedule, a reinforcer is always provided after a fixed interval of time. This can be specified on line 40 of the FT program. The VT schedule provides reinforcement at variable intervals, specified by a minimum and maximum interval on lines 41–24 of the VT program. These types of schedules are known for producing superstitious behavior, as the presentation of the reinforcer may strengthen some behavior that is completely unrelated to producing the reinforcer.
We also included fixed duration (FD) and variable duration (VD) schedule of reinforcement programs. These schedules have been less common in operant laboratories often because other experiment controllers make recording response duration difficult. Our software, however, records the duration of every event in the background, allowing for duration-related contingencies to be easily created. In our FD and VD programs, reinforcement is only provided when a single response lasts long enough to meet the duration requirement. Individual response durations do not sum to meet the requirements. For example, in an FD 2 second schedule, a rat must press and hold the lever for 2 seconds before receiving reinforcement; however, two responses of 1 second each will not produce reinforcement. A good way to explore FD and VD schedules is to run several sessions of continuous reinforcement, then have the students review the data spreadsheets to find the average and maximum response durations. Set the duration requirement (line 42 for the FD program, and lines 43–44 for the VD program) to the maximum response duration previously recorded. How does this change the response duration? Does response topography also change? See how long a response duration the students can train.
While FR and VR schedules require an increase in number of responses compared to continuous reinforcement, often resulting in an increase in response rate, differential reinforcement of high rate (DRH) schedules explicitly require an increase in response rate. In a DRH schedule, the subject is required to make a fixed number of responses within a fixed interval of time. In this sense a DRH can be considered an FR schedule with an additional time limit requirement. On line 46 of the DRH program, you can set the response requirement, or the number of responses that must occur to produce reinforcement. On line 47, you can set the interval requirement in milliseconds. In order for reinforcement to be produced, all responses must occur within the specified interval. For example, if the user requires 10 responses within 5 seconds, reinforcement will only be provided if they all occurred within the last 5 seconds. One interesting classroom exercise would be to compare performance on FR, VR, and DRH schedules to see which schedules are most successful at generating high response rate.
Differential reinforcement of low rate (DRL) schedules are designed to produce very low rates of response. Like FI schedules, in DRL schedules reinforcement is provided when a response occurs after a fixed time interval elapses. Unlike FI schedules, any response that occurs before the interval elapses resets the interval. For example, in a DRL 5-minute schedule, reinforcement will be provided for a response that occurs 6 minutes after the last reinforcer. However, a response that occurs 4 minutes after the last reinforcer will cause the interval to reset. The subject will now have to wait an additional 5 minutes before responding will produce reinforcement. You can specify the length of the response interval on line 42 of the DRL program.
Our final program is a simple extinction program. The program monitors any responses, but reinforcement is never provided. The extinction program can be used after stable behavior is established in another program to reveal additional effects of that contingency. For example, responding on VR schedules are more resistant to extinction than continuous reinforcement.
Although these experiments were created with traditional pigeon and rat laboratories in mind, they can be conducted with a much greater variety of stimuli, reinforcers, responses, and species. Invertebrates again make convenient and capable subjects. Free-flying honeybees will respond for sucrose solutions as reinforcers. Part or all of these procedures can be automated. See Abramson (1990) for some un-automated exercises and see Sokolowski and Abramson (2010) for a description of a fully automated apparatus. Crabs also respond well in operant contingencies and can learn to press a microswitch lever for a liquid food reinforcer (Abramson, 1990; Abramson & Feinman, 1990). The food can either be pumped to the crab manually to maintain students' attention, or provided automatically through an inexpensive peristaltic dosing pump for an aquarium.
Non-traditional vertebrates also make good subjects. Many species of fish can be trained using a similar method of reinforcement delivery as the crab. Fish also respond well to both LEDs and vibrating motors as stimuli (Miskovksy, Becker, Hilker, & Abramson, 2010). Some reptile species may also be suitable for classroom use. Snakes have been conditioned to press microswitches for a small amount of water reinforcement dispensed through an electronically controlled valve (Kleinginna, 1970; Kleinginna & Currie, 1979). Several designs exist for inexpensive apparatuses to study learning in aquatic turtles by providing food reinforcement (Bitterman, 1964; Pritz, Bass, & Northcutt, 1973). Heat may also be a good reinforcer for many reptile species (Kemp, 1969) and can easily be provided using a relay to turn on and off a basking lamp.
For student operant exercises, a pushbutton as a response device and an LED as a reinforcer can allow students to experience first-hand the power of schedules of reinforcement. If the students are instructed that the LEDs are reinforcers, they will typically try to earn as many LED reinforcers as possible. Students may be surprised to find that graphs of their responses closely resemble the classic graphs produced by pigeons and rats. See the video tutorials for an example of setting up a simple operant experiment with a student.
Discussion
We have found the controller to be reliable, easy to use, and extremely portable. The controller allows both the instructor and student to conduct fully or partially automated experiments depending on the need of the assignment. For example, students in our laboratory have used this controller to conduct research on habituation of the glowing response of bio-luminescent algae, the ability of planarians to seek water, and the social preferences of pigeons. These types of projects would be extremely difficult if performed manually. Many would not be possible with commercial experiment controllers. Instead of restricting the kind of research that can be conducted, as do many commercial devices, the Propeller Experiment Controller allows for a much greater range of studies to be conducted, encouraging a comparative approach.
For colleges that have no animal teaching laboratories, the Propeller Experiment Controller allows the possibility of a variety of in-home projects. Students can use their own pets to explore principles of behavior analysis and comparative psychology. For example, with just a pedal and a feeder device attached to the controller, a dog can easily be trained to press a pedal under various schedules of reinforcement. A slightly modified fish stick (Miskovksy, et al., 2010) can also be easily attached to the controller for fully automated experiments with fish or other aquatic species.
An additional benefit of this device is that students become re-acquainted with apparatus design. The ability to interface an apparatus with a controller is becoming a lost skill at both the undergraduate and graduate level. Our experiment controller simplifies many of the technical aspects of creating programs for experiments, and provides an easy introduction into electronics and programming skills. Additionally, the controller is so inexpensive that any student can afford to purchase one themselves, or a school-owned device can be lent to them with little financial risk. Neither option is possible with expensive commercial equipment, further preventing students from acquiring these useful skills. The knowledge that doctoral students gain working with the Propeller Experiment Controller is also of great benefit after graduation, as these skills and the device itself can be extremely useful for a new faculty starting a laboratory with a tight budget.
Finally, the need for flexible, inexpensive equipment is especially notable in countries with less developed behavioral psychology disciplines. Behavioral teaching laboratories are noticeably absent in developing countries. This is due in part because of the high cost of teaching-related equipment (Abramson & Bartoszeck, 2006). A lack of affordable equipment for researchers and classroom demonstrations means that the study of behavior is not able to thrive as a science in these areas. For example, previous research has found that many students in northeastern Brazil do not consider psychology to be a true science, in part because few universities contain laboratories devoted to research or teaching of psychological principles (Morales, Abramson, Nain, Junior, & Bartoszeck, 2005). The Propeller Experiment Controller presents a unique solution to this problem as both research and teaching laboratories can be created with minimal financial investment. Free software, such as Google Translate (Google Inc.; Mountain View, California), makes it possible to instantly translate portions of this paper and comments in the programs into other languages. Although these translations are not perfect, they offer an easy first step toward developing teaching laboratories in other countries.