Appetite vs. Consummation in Operant Responding
I had a vigorous discussion with a colleague yesterday on what constitutes an appetitive phase (App) vs. the consummatory phase (Con) of a behavior. To provide some specific context, the issue was the pharamacological effect of 5HT or DA drugs on taste and feeding behavior, and whether the effects could be experimentally determined to be restricted to acting on the App vs. Con phases. The data in the talk were very nice, and the presentation was excellent: clearly 5HT and DA modulate food intake in different ways, and 5HT appears to be critical in modulating taste perception and palatability rather than motiviaiton to intitiate or maitain a meal. But I took a pedantic interest in the use of App and Con as descriptors — and I think they are mis-applied in the behavior of operant responding.
The 3 examples of mapping App and Con onto behaviors were
- initialiating a sucrose meal: approach to a spout (App) vs. licking at the spout (Con)
- licking within a sucrose meal: the end of an interburst interval (0.5 s pause?), relected in # of bursts initiated (App) vs. licking within a burst, relected in rate of licking (Con)
- operant feeding for sucrose: dry licking on an operant spout, equivalent to bar pressing, in a progressive ratio (App) vs. amount of sucrose consumed, reflected in # of rewards received (Con)
I dispute all of these mappings interpretations, but I focused on the operatant responding. It seemed to me that operants are usually trained up to become stereotyped, ballistic behaviors in response to a cue (like the cue light coming on, or the bar being made available). Thus waiting and looking for the cue is the App, which triggers the operant bar pressing which is itself the Con. The reception and consumption of the reward is an independent behavioral sequence, and so don’t enter into it.
First, just to get it out of the way: even though the example behavior is feeding, the “consummatory” of Con is “consummation”, not “consumption”. So many behavioral sequences have a Con phase, such as mating, or prey acqusition, and probably lots of other things like pup retrieval, grooming, etc.
I have to concede the point that it is a convention in the behavioral neuroscience of operant conditioning that the operant response is called the App, and the reception of the reward is called the Con. For example, in bar-pressing for access to a female rat, the male’s bar-pressing is called the App, and the act of mating is called the Con. So it appears to be a convention, and the convention can be applied to dry-licking for a lick-mediated sucrose reward.
But I like the original ethological definition of App vs. Con. To cite the canonical example from Tinbergen: a hawk meanders above a field until it sees a mouse, then it enters a dive to apprehend the mouse. The whole behavioral sequence has the function of hunting and acquiring prey. The meandering is the App, the dive is the Con (because “diving to latch onto the mouse” is the behavior which consummates the “acquiring prey” sequence). I believe Tinbergen defines the phases as: the App phase is characterized by non-stereotyped “random” movements that do not rely on sensory perception of the target (the hawk will fly in a different pattern each day), but the App phase ends up bringing the target within sensory range. One the target is preceived,preception of it acts as a “releaser” of the Con behavior, which is a very stereotyped and ballastic behavior (in the case of the hawk’s dive, literally ballistic.)
So the App is not stereotyped, and not in direct sensory response to a stimulus. To me, bar-pressing seems very stereotyped and is triggered by perception of the cue-light or the bar itself; indeed, the act of pressing the bar consummates the behavior of the well-trained operant rat.
Alan offered a different interpretation, which is that bar-pressing is the act which brings the rat into sensory reception with the reward stimulus (access to sucrose), and that the licking of the sucrose is Con (I guess because it consummates the behavior of ingestion.) This is similar to the schema in which a Con is required essentially as a reflex sensory-motor response: given sucrose on the tongue, the rat will reflexively generate licks and other orofacial movements to drive ingestion. So the Con is not just stereotyped and ballastic, but it is down-right reflexive (and I detect a hint of non-volitional as well).
(I still like my counter example of bar pressing for brain self-stimulation: after bar-pressing, no behavior is required to consummate the application of current to the reward pathways. So what is the Con in brain self-stimulaton responding? )
I myself have used this definition in distinguishing the behaviors driven by ad lib bottle access vs. intraoral infusions. The bottle requires App for the rat to find and approach the bottle, and the Con is licking at the bottle. An intraoral infusion involves no App, because the rat is just sitting there when the infusion starts. Sucrose magically appears in its mouth, so the rat licks and swallows: a Con with no App.
The bottom line, and then a hypothesis. The bottom line is that i) Clare and Alan’s definition of bar-pressing as App is conventional and legitimate, but ii) I think it is a misinterpretation by simplification of the ethological definition of Tinbergen, although iii) it can all be resolved by admitting that one person’s App is another person’s Con. Barpressing is the Con of the “see cue light, start pressing bar” sequence, but it is the App of the “press bar to get access to sucrose” sequence.
App == Limb Movement
A hypothesis: the behavioral neuroscience usage of App and Con usually relies on very indirect definitions, based on the variable being recorded, rather than on direct observation of the whole animal behavior. Depressions of a bar are interpreted as App, Long interburst intervals are interpreted as App (i.e. a gap in data collection!) A common factor in many so-called Apps is the non-reflexive movement of the limbs.
I base this on an observation made by John D. Davis many years ago, while showing a video of a rat in a lickometer. Jack divides pauses into interburst intervals (dividing licks into bursts) and longer intercluster intervals (dividing clusters of bursts). IBIs are on the order of 0.5 s, but ICIs are on the order of seconds. Jack casually pointed out that during IBIs, the rat stays with its head positioned at the lick spout, and just pauses by skipping a few licks. During ICIs, however, he thought that rats often moved their hindquarters, so that the head left licking position and needed to be approach and re-orient to resume licking. Although I don’t think anyone has ever quantified this, it seems almost as if descending volitional movement is suspended within a cluster, and that locomotion or limb movement interrupts the reflexive licking. So perhaps what neuroscientists call App is behaviors that involve volitonal movement of the limbs? This would map onto, e.g., walking to the bottle, or fore-limb bar pressing. Furthermore, one could start to hang a neurological scheme onto this behavioral scaffold (App == cortex & basal ganglia inititated behaviors?)
[Aside] Does a progressive ratio have to be progressive? The typical progression is somethin like: 10, 50, 100, 150, 200, 250, in an ascending series. What would happen, if you train a rat to barpress close to the breakpoint for a stimulus (e.g. 100 presses /pellet), and then present different ratios in random order (i.e. 50, 100, 20, 60, 200, 75, 10, 90…). Would the rat still give up and not bar press at the supra-breakpoint values, or would the apparent breakpoint be higher or lower? I wonder if the progressive ratio depends on a memory of previous effort, and internally the rat recognizes the upward slope of the effort. What if there was no consistent slope, would it still reveal the same effort? (so “motivation” ∝ changes (slope) of recent efforts, not the current effort per se.)[end of Aside]