Co-authored by Dr. Traci Cihon, Associate Professor, University of North Texas
There are always outdated notions and misconceptions about any area of research. As scientists conduct more research and historians uncover more information, we can correct these misconceptions. Unfortunately, not all psychology textbook authors have updated their information accordingly, and instructors are still incorrectly teaching students some concepts. Even if it is possible to update textbook information after an edition has been published (see, e.g., Greene, 2019, April 12), it would be better to ensure that students don’t need to convince publishers or authors to do so. We’ll cover three such misconceptions about Pavlovian conditioning that you may know better as classical or respondent conditioning: 1) what produces learning and relatedly what Pavlov called the stimuli involved in classical conditioning, 2) what stimuli Pavlov used in his experiments, and 3) what happened to Little Albert.
In a simple example of excitatory conditioning, we might start with an unconditional reflex that occurs without learning. For instance, your dog will salivate when he eats his food. Food is the unconditional stimulus that will elicit salivation as the unconditional response (but see Domjan, 2016 for a discussion on elicited versus emitted behavior). You can’t prevent salivation from occurring when he’s eating, and you wouldn’t want to given that salivation is the first step in digestion. All animals, including people, pick up on signals from our surroundings to help us predict what will happen next. Perhaps when you’re about to feed your dog, you bring out his bowl and the bag of dog food. The dog food bag and empty bowl are neutral stimuli that can become effective signals for food if they reliably predict that the dog will get to eat. If you get the empty bowl and set the dog food bag down next to the bowl just before you feed your dog twice a day for a week, you may notice that your dog salivates when he sees the bowl and the bag. The empty bowl and dog food bag have become conditional stimuli that tell your dog when his next meal will occur. You know this because he salivates not only to the food but even before he’s eating — when he sees the bowl and the bag of food; this salivation in anticipation of the food is the conditional response. The conditional reflex appears after learning/conditioning has occurred, and an otherwise meaningless stimulus (e.g., food bowl and bag of food) becomes a predictor of food. We want to know when food will be available because it is a biologically important event (see also Eelen, 2018). This is the type of digestion scenario that most students know as Pavlovian conditioning because the unconditional stimulus occurs quickly after the conditional stimulus (i.e., short-delay conditioning). Notably, Pavlovian conditioning is also responsible for fear conditioning (Fullana et al., 2016; Sehlmeyer et al., 2009) and evaluative conditioning (De Houwer et al., 2001; Gast et al., 2012; Hofmann et al., 2010) in people.
Unconditional reflex occurs in Step 1, the neutral stimulus reliably signals the unconditional stimulus in Step 2, and the neutral stimulus becomes a conditional stimulus in Step 3 which elicits the conditional response
The first misconception is that Pavlovian conditioning occurs because events are paired. As an example, Hockenbury and Hockenbury (1999) said, “Pavlov repeatedly paired a sound (the initially neutral stimulus) with food (the unconditioned stimulus, or UCS), which reflexively triggered salivation (the unconditioned response, or UCR). After experiencing this pairing several times, the dogs became classically conditioned” (p. 336). Specifically, the pairing between the conditional stimulus (e.g., empty food bowl) and the unconditional stimulus (e.g., food) is the necessary and sufficient condition for learning. Typically, interstimulus intervals (i.e., the time between the onset of the conditional stimulus and the unconditional stimulus) shorter than 4 s are better for conditioning responses like eyeblink, muscle contraction, and siphon withdrawal (e.g., Hawkins et al., 1986; Lederhendler & Alkon, 1989; Steinmetz et al., 2011; Woodruff-Pak et al., 2007) in nonhuman animals, but people can tolerate slightly longer interstimulus intervals (cf., Kjell et al., 2018). This does seem to support the assertion that pairing the conditional stimulus and unconditional stimulus produces the conditional response. However, pairing these stimuli or having them occur close together in time is not required for learning. Learning can occur when the conditional stimulus and unconditional stimulus are separated by hours or minutes (i.e., unpaired as in trace conditioning) with conditioned taste aversions for people and nonhuman animals, respectively (e.g., Chambers, 2018; Houpt et al., 2015; Lin et al., 2017). It wouldn’t make sense to elicit rapid blinking minutes before you encountered an eye irritant (cf. Cheng et al., 2008), but it does make evolutionary sense to feel nauseated when you see or smell some moldy food and avoid eating something that might poison you (cf., Reilly & Schachtman, 2009; Schier et al., 2019; Sugai et al., 2006). Cancer patients, for example, undergoing treatment will develop aversions to novel flavors consumed before chemotherapy (e.g., Berstein, 1985; Jcobsen et al., 1993; Mattes & Boraas, 1987; Scalera & Bavieri, 2009) when the novel flavor conditional stimulus and chemotherapy unconditional stimulus are separated in time. Coincidentally, learning doesn’t always occur when the conditional stimulus and unconditional stimulus are presented at the same time as in simultaneous conditioning (e.g., Bangasser et al., 2006) or backward conditioning (e.g., Mahoney & Ayres, 1976). As Eelen (2018) stated, “[I]t would be maladaptive for an organism if the mere coincident occurrence of two events would be a sufficient condition for the organism to establish a connection between the [conditional stimulus and the unconditional stimulus]” (pp. 197-198).
Part of the pairing misconception is that Pavlov called these paired events conditioned and unconditioned stimuli. Rescorla (1988) also pointed out that Morgan and King (1966, pp. 79-80), Klatsky (1980, p. 281), Gardner (1982, p. 594), and Rosenhan and Seligman (1984, p. 669) expressed similar sentiments about the pairing of the conditioned stimulus with the unconditioned stimulus. If we want to be consistent with Pavlov’s terms and findings, we should refer to those events as conditional and unconditional stimuli. Trow (1929) stated, “The behaviorist’s most cherished possession is the conditioned reflex (the Russian word is conditional), furnished by the researches of Dr. Pavlov” (p. 275). Specter (2014, November 24) also stated, “Pavlov is perhaps best known for introducing the idea of the conditioned reflex, although Todes notes that he never used that term. It was a bad translation of the Russian uslovnyi, or “conditional” reflex. For Pavlov, the emphasis fell on the contingent, provisional nature of the association — which enlisted other reflexes he believed to be natural and unvarying.” The emphasis on conditioned events simply means that learning has occurred. The emphasis on conditional events means that we’re considering the probability of one event given that another has occurred; we’re pointing to the contingency between the conditional stimulus and the unconditional stimulus rather than the fact that they were presented contiguously (e.g., Dinsmoor, 2004). Thus, we communicate what we think are the necessary and sufficient conditions for classical conditioning by the way that we talk about those events, and we perpetuate misconceptions about pairing while ignoring contingency when we favor the mistranslation of Pavlov’s original sentiments.
Rescorla (1968, 1969, 1988) has shown that conditioning occurs when a conditional stimulus signals a change in the probability of encountering an unconditional stimulus. In all conditions for at least some proportion of trials, the tone conditional stimulus and the shock unconditional stimulus were paired. Rats only showed the fear conditional response to the tone when the tone signaled that shock was more likely. That is, the probability of shock after a tone was higher than the probability of shock after no tone. When the probability of shock after a tone was equal to or lower than the probability of shock after no tone, rats were not afraid of the tone. Although stimulus contiguity is often assumed when there is a contingency between events, the contingency – and not pairing – is the necessary and sufficient condition for learning via Pavlovian conditioning (cf., Balsam et al., 2010; Eelen, 2018; Jenkins & Shattuck, 1981; Kitaguchi, 2000; Tanner et al., 1987; Witnauer & Miller, 2008).
I. P. Pavlov
The second misconception about Pavlovian conditioning is that Pavlov didn’t use a bell, but students in introductory psychology courses typically hear about how Pavlov used a bell as a conditional stimulus to signal that dogs in his original experiments would get food. Catania (1994), Black (2003), Littman (1994), and others have expressed doubt as to whether Pavlov used a bell in his research presumably because the sound of a bell can be difficult to control with experimental precision. Despite claims to the contrary, Thomas (1997, p. 116), Jarius and Wildemann (2015), Tully (2003a, 2003b), and Rehman et al. (2019) have documented that Pavlov did use a bell or buzzer. Pavlov used many different stimuli as conditional stimuli once he realized that any stimulus, including the sight of a laboratory assistant (Eelen, 2018, p. 197), that reliably preceded an unconditional stimulus could elicit a conditional response. In point of fact, the first neutral stimulus that become a conditional stimulus in Pavlov’s and Vul’fson’s studies on digestion was the sight of food presented at a distance before the dogs were fed the meat powder, which effectively gave scientists a way to systematically study the effects of the environment on objective behavior (Gantt, 1973; Windholz, 1989). Previously, psychologists relied upon subjective accounts of behavior (Windholz, 1986). It was this difference in Pavlov’s approach to studying physiological and psychological responses to environmental stimuli that most influenced Skinner (Skinner, 1996) and thus behavior analysis specifically (Dinsmoor, 2004), behaviorism generally (Windholz, 1983), and neuroscience (see Pickenhain, 1999).
The third misconception about Pavlovian conditioning is that Douglas Merritte was Watson and Rayner’s Little Albert. Little Albert was the 11-month-old boy who participated in Watson and Rayner’s (1920) fear conditioning experiment (Griggs, 2014). Most people seem to dislike Watson (and behaviorism; see for example, Malone & García-Penagos, 2014) due to his popular quote which is typically included out of context (Rakos, 2013, p. 102) and was published at a time when the nature-nurture debate raged on with most psychologists heavily favoring genetic explanations for behavior.
Give me a dozen healthy infants, well-formed, and my own special world to bring them up in and I’ll guarantee to take any one at random and train him to become any type of specialist I might select – doctor, lawyer, artist, merchant-chief, and yes, even beggar-man and thief, regardless of his talents, penchants, tendencies, abilities, vocations, and race of his ancestors. I am going beyond my facts and I admit it, but so have the advocates of the contrary and they have been doing it for many thousands of years.
Watson (1924, p. 82; 1930, p. 104)
In the fear conditioning experiment, Albert B. was pretested to see what his reaction would be to several stimuli, including a white rat, a rabbit, a dog, and masks with and without hair. He wasn’t afraid of any stimuli that he saw for the first time. Albert startled the first two times that the experimenters hit a steel bar to produce a loud sound but cried after the third time (Watson & Rayner, 1920). To condition fear, they produced the loud sound after Albert started to reach for the white rat. Then Albert associated the loud sound unconditional stimulus with the white rat conditional stimulus and would startle in the presence of the white rat. This fear generalized to the rabbit, the fur coat, and the Santa Claus mask on their first presentation after conditioning with the white rat but not the dog, the blocks, or the wool. With additional fear conditioning, this fear response lasted for longer than a month. Albert would also attempt to move away from the conditional stimuli, but this was an operant avoidance response. Watson and Rayner (1920) had a tentative plan to extinguish Albert’s fear of fuzzy objects, but he left the hospital before they attempted systematic desensitization. While this experiment wasn’t as rigorous of a demonstration of experimental control as Pavlov’s studies (but see Paul & Blumenthal, 1989), it was the first empirical demonstration of acquired (but transient) fear in a person.
Little Albert next to J. B. Watson Rosalie Rayner
Prompted by sensationalist narratives of Little Albert being permanently harmed by Watson, and by extension, behaviorism, the search began for the fate of Albert B. Harris (2011) inadvertently renewed this interest in searching for Albert despite the fact that he was more interested in the history, mythology, and years of research that were influenced by classic experiments like Watson and Rayner’s (1920) conditioned emotional responses. He ultimately reasoned that it would be difficult if not impossible to find the actual participant behind Albert B. The most popular story is that Albert B. was Douglas Merritte who died as a young child from hydrocephalus (Beck et al., 2009; Fridlund et al., 2012), although Albert didn’t appear to have any neurological impairments in Watson’s footage. The patient records from Johns Hopkins University were lost, so Beck et al. (2009) couldn’t confirm that Merritte was (a)neurotypical. Alternatively, Powell et al. (2014), Digdon et al. (2014), and Griggs (2015) proposed that Albert Barger was the identity of Little Albert. Barger, unlike Merritte, lived a long and relatively normal life. Coincidentally, Barger didn’t have any specific phobias, but he didn’t particularly like dogs. Regardless of the fate of the actual Albert B., fear conditioning has been an active area of research and can still provide useful information for practitioners who want to treat specific phobias (Mertens et al., 2020; Ollendick & Muris, 2015).
Hopefully by correcting these misconceptions in textbooks about how Pavlovian conditioning occurs and how early researchers conducted their original experiments, future generations of students will appropriately understand how learning affects behavior. They may even revive formerly unresolved lines of research in Pavlovian conditioning or discover new questions that need answers. For instance, is Domjan (2016) correct in saying that operant conditioning and classical conditioning have very few, if any, differences given that behavior analysts also study antecedent stimulus-consequence relations (e.g., stimulus pairing in derived stimulus relations; Dougher et al., 1994; Minster et al., 2011; Roche & Barnes, 1997). Is there a better model than the Rescorla-Wagner model that can account for the many Pavlovian conditioning effects (Bitterman, 2006; Miller et al., 1995)? Alternatively, we could answer a popular question posed by students: What are the ideal interstimulus interval ranges for each type of unconditional response (e.g., eyeblink, fear, taste aversions, salivation, blood pressure/heart rate changes) to produce robust conditional responses?
Image credits:
- Image provided courtesy of pxhere under CC0 Public Domain
- Image provided courtesy of pxhere under CC0 Public Domain
- a. Image provided courtesy of Mart Production under Pexels License, b. Image provided courtesy of Helena Lopes under Pexels License, and c. Image provided courtesy of Mart Production under Pexels License
- Image provided courtesy of Tretyakov Gallery under Public Domain
- Image provided courtesy of Akron psychology archives under Public Domain