Comments by Anthony Sclafani (February, 2016)

In what I consider a “mini” classic because it is not frequently cited, Garvin L. Holman (1968) reported in a paper entitled “Intragastric reinforcement effect” that rats learned to prefer a flavored non-caloric solution that was paired with an intragastric (IG) infusion of a nutritive diet. This appears to be the first report of a post-oral nutritive-conditioned flavor preference (CFP). An earlier paper by Le Magnen (1959; 1999) is sometimes cited as demonstrating glucose-induced CFP, but he actually reported that an intraperitoneal injection of 30% glucose conditioned a flavor avoidance. Holman trained hungry rats to drink a flavored saccharin solution (referred to here as the CS+) for 5 min followed immediately by an IG infusion of a complex eggnog diet and on alternate days to drink a differently flavored saccharin solution (the CS-) paired with an IG water infusion. The day after the last training session, a two-bottle choice test (140 min) was conducted with the CS+ vs. CS- solutions without IG infusions. The rats showed a significant CS+ preference during the first 20 min of the test which then dissipated during the remaining 120 min of testing. Although the preference was modest (~65%) and temporary, perhaps because of the short training sessions and complex diet used, Holman’s findings established that IG nutrients can reinforce a flavor preference.

Over the next 20 years several more studies reported that IG infusions of various nutrients (milk, glucose, casein) conditioned flavor preferences in hungry rats (Puerto et al. 1976; Mather et al. 1978; Sherman et al. 1983; Baker et al. 1987). Given the deprivation state of the animal, it was assumed that nutrient restoration was the effective “reinforcer” for the conditioned flavor preference. Holman himself presented his study as a test of the “need reduction” (or drive reduction) hypothesis made famous by Clark Hull (1943). However, my lab reported in 1988 that rats with unlimited access to a nutritionally-complete chow acquired a very strong (96%) preference for a flavored solution paired with concurrent IG self-infusions of a maltodextrin solution during 24 h/day sessions (Sclafani and Nissenbaum 1988). Furthermore, this preference persisted for 4 to 40 days of extinction testing in which the CS+ and CS- flavors were paired with water or no infusions (Sclafani and Nissenbaum 1988; Elizalde and Sclafani 1990). Non-deprived as well as food deprived mice also learn to prefer CS+ flavors paired with IG self-infusions of glucose or fat (Sclafani and Glendinning 2005; Sclafani and Ackroff 2012; Zukerman et al. 2011). Energy repletion may contribute to the flavor conditioning produced by IG nutrient infusions in hungry animals, but it is not essential. Hungry mice learned to prefer a CS+ flavor paired with IG infusions of a non-metabolizable glucose analog that provided no energy but failed to prefer a CS+ flavor paired with IG infusions of a caloric fructose solution (Zukerman et al. 2013b). These and other findings implicated the intestinal sodium-glucose transporter 1 (SGLT1), which binds to glucose but not fructose, as a critical sensor mediating glucose-conditioned flavor preferences (Sclafani et al. 2016; Zukerman et al. 2013a). Other findings indicate a role for hepatic-portal glucose sensing in preference conditioning (Oliveira-Maia et al. 2011; Tordoff and Friedman 1986). Intestinal GPR40 and GPR120 fatty acid sensors, in contrast, are implicated in flavor conditioning by IG fat infusions (Sclafani et al. 2013).

Holman’s IG flavor conditioning experiment was actually the third experiment in this seminal publication. The first two experiments focused on the ability of IG nutrient infusions to reinforce bar pressing behavior in rats. Prior studies by Epstein and Teitelbaum (1962) appeared to establish that IG nutrient infusions could reinforce bar press responding in rats. In preliminary studies, however, Holman failed to replicate their findings, which led to his systematic analysis of the “intragastric reinforcement effect.” In brief, Holman reported that rats would bar press for IG nutrient infusions only if the infusions were associated with orosensory stimulation. The original experiments of Epstein and Teitelbaum used a perishable liquid diet that was refrigerated during the training sessions. The diet was infused through a nasogastric tube, which provided thermal stimulation of the oropharyngeal region. Holman confirmed that rats would bar press for cold nasogastric diet infusions. However, rats tested with a warm liquid diet infused via a nasogastric catheter or a diet infused through a subcutaneous IG catheter that bypassed the oropharyngeal region were much less likely to bar press for IG nourishment. A few rats successfully fed themselves IG in these later test conditions but Holman observed that they “looked as if they were ‘eating’ the bar” and “seemed to rely heavily on self-produced oral stimuli derived from licking and gnawing” the bar. Rats tested with a retractable bar, which minimized bar licking, did not respond for IG nutrient infusions. If the IG infusion was paired with presentation of a dilute saccharin solution, which by itself did not support bar pressing, then the rats would bar press for the saccharin plus IG nutrient reinforcement. Holman concluded that it was very unlikely that IG nutrient infusions reinforce operant bar pressing directly but rather modulate the reinforcing value of the oral sensations associated with the nutrient infusion.

Following Holman’s paper, Cytawa et al. (1972) reported that naïve rats did not learn to bar press for IG infusions of an eggnog diet, but after being trained to bar press for oral presentation of the diet, the same rats bar pressed for IG diet infusions even though the diet was infused at body temperature via a nasogastric catheter. The authors did not comment on whether the rats licked or gnawed the bar during the infusions. A subsequent paper by Trojniar and Cytawa (1976) reported that rats switched from oral diet delivery to IG diet infusions showed vigorous bar pressing in the first IG session. However, like rats infused with IG water or nothing, the diet-infused rats showed a decline in bar pressing (i.e., extinction) over the next nine test sessions. Interestingly, rats infused with IG morphine rather than liquid diet displayed sustained bar pressing over the ten extinction sessions. These 40-year old findings have implications for the current discussion on the “addictive” qualities of food and particularly claims that food can be as or more addictive than some drugs of abuse (Ahmed et al. 2013). It may be, however, that the eggnog diet is not the most effective food to support bar pressing for IG infusions, and future studies should investigate different high-sugar or high-fat food sources (Avena et al. 2008; Tellez et al. 2013a).

In a recent series of innovative studies, de Araujo and colleagues (2012; 2013b; 2013a) have revived the operant IG feeding paradigm, although in this case using mice rather than rats and spout licking rather than bar pressing as the operant. Hungry mice were first induced to lick a stainless steel spout containing a food odor cue, with each reinforced lick paired with a small IG infusion of Intralipid (soybean oil emulsion) via a subcutaneous catheter. After four 1 h/day training sessions, the mice continued to lick when offered an odorless dry spout, and their reinforced licking responses varied as a function of lipid concentration (7.5%, 15%, or 30%) so as to maintain relatively constant 1-h caloric intakes. Ferreira et al. (2012) reported similar findings with mice operant licking for IG glucose infusions. My lab confirmed that IG Intralipid and glucose infusions reinforce operant licking in mice (Sclafani et al. 2015; Sclafani and Ackroff 2016). Ferreira et al. (2012) concluded that “animals are capable of rapidly and precisely regulating calorie intake in the absence of oropharyngeal sensations throughout the entire feeding episode” [emphasis added]. However, Holman would have argued that dry spout licking provided orosensory stimulation for the mouse, just as bar licking and gnawing did for rats bar pressing for IG nutrient infusions. With respect to caloric regulation, note that while mice will operant lick for IG glucose infusions they fail to do so for isocaloric fructose infusions (Sclafani and Ackroff 2016). It remains to be established if mice or rats would bar press (rather than spout-lick) for IG fat or glucose infusions if orosensory stimulation is minimized using a retractable bar as in the Holman 1968 study. Perhaps (or perhaps not) there is a special relationship between orosensory stimulation and post-oral nutrient reinforcement as proposed by Holman (Holman 1968) and others (Garcia and Koelling 1966). The IG fat and sugar infusion parameters established by the de Araujo lab are ideal to test this notion given the vigorous operant licking behavior they produce in mice.

References