Is the FEO food entrainable? - Reexamining the classic Fred Stephan experiment using canonical-clock–less Period 1/2/3 triple knockout mice

Abstract

When nocturnal rodents are subjected to daytime restricted feeding, in which food is only available for a few hours per day, they typically become active a few hours before the onset of the scheduled mealtime. This so-called food-anticipatory activity (FAA) is controlled by an autonomous circadian pacemaker, which is independent from the central circadian pacemaker in the suprachiasmatic nucleus (SCN). Fred Stephan named this pacemaker the food-entrainable oscillator (FEO) because FAA re-entrains to a shifted feeding schedule. We recently developed a method to measure food-seeking nose-poking behavior by an operant feeding device and found that anticipatory food-seeking nose-poking for scheduled daily food availability shifts in parallel with phase-shifted environmental light-dark cycles, raising the possibility that anticipatory food-seeking behavior is controlled by an oscillator entrained to the environmental light-dark cycle. With this possible light-entrainability of the FEO, we revisited Stephan’s historical experiment—testing whether the FEO entrains to feeding cycle in the absence of a light-dark cycle without functional SCN—using Period 1/2/3 triple knockout (KO) mice, in which the canonical circadian oscillators in the SCN and peripheral tissues are disabled. KO mice were subjected to restricted feeding under constant darkness. The food-seeking nose-poking activity of a subset of the KO mice indeed occasionally entrained to the feeding cycle and re-entrained to a shifted feeding cycle. Despite our previous study showing that anticipatory food-seeking behavior shifted with the environmental light-dark cycle, these data demonstrate that it can also entrain to the feeding cycle in the absence of an environmental light-dark cycle, supporting Stephan’s observation that the FEO is indeed food-entrainable.

INTRODUCTION

In 1922, Curt Richter reported his observation that spontaneous activity increases during the several hours before the time of daily restricted meals in rats kept in either constant darkness or constant light (Richter, 1922). In 1979, a series of experiments from Fred Stephan demonstrated that this so-called food-anticipatory activity (FAA) was controlled by an autonomous circadian oscillator that was located outside of the central primary circadian pacemaker in the suprachiasmatic nucleus (SCN) (Boulos et al., 1980; Boulos and Terman, 1980; Stephan et al., 1979a; Stephan et al., 1979b). Stephan named this oscillator the food-entrainable oscillator (FEO) because FAA in SCN-lesioned rats was able to entrain to the feeding cycle under constant dark conditions, in the absence of light-dark time cue (Stephan, 2002).

Several lines of evidence have demonstrated that the FEO has the characteristics of an autonomous circadian oscillator (Stephan, 1984; Stephan, 1986a; Stephan, 1986b; Stephan, 1986c; Stephan, 1992; Stephan, 2001; Stephan, 2002; Stephan et al., 1979a; Stephan et al., 1979b). SCN-lesioned rats exhibited FAA in constant darkness. The FAA seen in SCN-lesioned rats showed transients during re-entrainment to phase-shifted feeding times. Additionally, during restricted feeding schedules outside of the range of entrainment (restricted feeding at a 23h interval, T23 feeding), FAA exhibited a rhythm with a period longer than 25h. Since Stephan first reported the FAA rhythm did not entrain to feeding cycle in 1992 (Stephan, 1992), two other studies have reported similar observations. We observed a short-period (~21h) activity rhythm in Period 1/2/3 triple knockout (Per1/2/3 KO) mice during restricted feeding at a 24h interval (T24) (Pendergast et al., 2012). Takasu and her colleagues reported that 1 of 7 SCN-lesioned Cryptochrome 1 knockout mice exhibited a ~22h period rhythm during 11 cycles of restricted feeding at a 21h interval (T21) (Takasu et al., 2012). It is worth noting that FAA rhythm with period different from the feeding cycle period is rarely observed, and FAA immediately disappears upon beginning ad libitum feeding. As a result, researchers often have to perform food deprivation subsequent to restricted feeding to show that the FAA reappears at the same time in the absence of the feeding cycle. This approach has been used as a standard protocol to demonstrate that FAA is controlled by an endogenous circadian pacemaker (Boulos and Terman, 1980; Pendergast et al., 2012; Stephan et al., 1979b). The difficulty of measuring the free-running rhythm of the FEO makes it challenging for researchers to use traditional lesion or forward/reverse genetic approaches to determine FEO candidate brain loci or genes involved in FEO timekeeping, respectively (Pendergast et al., 2012).

Another unique characteristic of the FEO is that its oscillation is independent of the canonical circadian genes. When experiments are conducted under a normal light-dark cycle (T24), circadian mutant mice exhibit FAA during restricted feeding, suggesting that the FEO is composed of a non-canonical molecular time-keeping mechanism that is independent of canonical circadian genes and coupled with a light-entrainable pacemaker (e.g., the SCN) (Iijima et al., 2005; Pendergast et al., 2009; Pitts et al., 2003; Storch and Weitz, 2009). This also makes infeasible the approach of testing circadian gene knockout/knockdown in subsets of cells to determine the location of the FEO.

Conventionally, FAA is assessed by measuring wheel-running activity, general cage activity, or core body temperature, but we utilized an open-source feeding device, the Feeding Experimentation Device 3 (FED3), that the Kravitz lab developed (Ehichioya et al., 2024; Ehichioya et al., 2023; Matikainen-Ankney et al., 2021). With this device, we measured a specific food-seeking nose-poking behavior. We set up the FED3 to dispense one pellet when the left nose-poke hole was poked. We have initially shown that when food availability is limited to a few hours during the day, C57BL/6N and C57BL/6J wildtype mice exhibit robust food-seeking anticipatory nose-poking a few hours before mealtime. Furthermore, with the FED3, food-anticipatory poking can be measured at any time of day, similar to in earlier pioneering studies measuring lever pressing for dispensing food (Boulos et al., 1980; Petersen et al., 2022; Petersen et al., 2014). This provided us with a reliable readout of the food-seeking behavior controlled by the FEO. Importantly, during restricted feeding, the FED3 only dispenses pellets during certain hours every day. There is no cue associated with feeding time. Therefore, mice have to either use an internal clock to predict the time when food is available or continuously or randomly poke until the feeding time arrives and poking dispenses pellets. By using Per1/2/3 KO mice, in which the canonical circadian oscillators are genetically disabled, we conducted several experiments that shifted mealtime and the light-dark cycle independently. We found that the time of anticipatory food-seeking behavior shifted in parallel with the light-dark cycle, suggesting that the FEO entrains to the light-dark cycle rather than to the feeding cycle (Ehichioya et al., 2023).

In the current study, we revisited Fred Stephan’s historical experiment in which he observed the food-entrainability of the FEO (Boulos et al., 1980; Stephan, 1992; Stephan et al., 1979a; Stephan et al., 1979b). We conducted experiments measuring the anticipatory food-seeking behavior of Per1/2/3 KO mice in both the presence and absence of the light-dark cycle. We found that the FEO of Per1/2/3 KO mice is indeed food-entrainable in the absence of the light-dark cycle.

Materials and Methods

Mice

Per1/2/3 triple KO mice maintained in C57BL/6J background strain (Ehichioya et al., 2023; Pendergast et al., 2012; Taufique et al., 2025) were used in this study. Some KO mice had one allele of the cfos-shEGFP transgene, which means the cfos-positive cells can be visualized with a short half-life of enhanced GFP (Reijmers et al., 2007; Taufique et al., 2025). However, we didn’t use this transgene in the current study. All procedures were carried out following the National Institutes of Health Guidelines regarding the care and use of animals for experimental procedures and were approved by the Institutional Animal Care and Use Committee at UT Southwestern Medical Center (Protocol #: 2016-10376-G). Both sexes were used in the current study. In the relatively small sample size used in the current study, no sex difference was noted. Sex of each individual has been indicated in the supplemental figures. All mouse care under constant darkness was done using an IR viewer, without exposing mice to visible light.

Wheel running, food-seeking behavior, and pellet intake measurements

Mice were individually housed in cages (32.5 cm in length, 14.5 cm in width, and 13 cm in height) equipped with an 11-cm diameter running wheel and an FED3 (LABmaker, Berlin, Germany) (Ehichioya et al., 2024; Matikainen-Ankney et al., 2021) with woodchip bedding (Sani-Chips, PJ Murphy Forest Products, Montville, NJ, USA). The cages were placed in light-tight ventilated cabinets. Temperature, humidity, and light intensity inside the cabinet were recorded every 5 min by Chamber Controller software (ver. 4.104, Actimetrics, Lafayette, IN, USA). Mice were under a 12h light (~200 lux at cage level, generated by white LEDs at the top of the light-tight box) and 12h dark condition or constant darkness controlled by the Chamber Controller software. Wheel-running activity (via micro-switches mounted to the sides of the running wheel cages) was continuously recorded every 1 minute by the ClockLab acquisition system (version 3.604, Actimetrics, Lafayette, IN, USA). Cages and water bottles were changed every 3 weeks. The FED3 was filled with 20 mg food pellets (F0163, Bos-Serv, Flemington, NJ, USA) and set to dispense one pellet by a single nose-poke to the left nose-poke hole. The right nose-poke hole was disabled; however, right nose-poking was recorded as exploratory activity. The FED3 was set to dispense the pellets only during certain times of the day during restricted feeding. The FED3 was set to not dispense any pellets during food deprivation. The times of pellet intake, right nose-poke, and left nose-poke events were recorded on the SD card inside the FED3. Pellets were filled every 3 days.

Experimental Procedures

Experiment I

Three male and 3 female Per1/2/3 triple KO mice (5.5–10 months old) were single-housed and maintained under a light-dark cycle with ad libitum access to food for 7 days. At ZT12 (time of lights-off), the day before restricted feeding started, mice were placed on an overnight fast for a duration of 16h. The next day, mice were subjected to a restricted feeding protocol with 8h of restricted feeding from ZT4–12 for 2 days, then 6h of restricted feeding from ZT4–10 for another 2 days. This was followed by 4h of restricted feeding (ZT4–8). To test the involvement of a light-entrainable circadian oscillator in food anticipation, mice were released to constant darkness at day 21. To test the role of hunger-induced food seeking in food anticipation (hourglass clock), unexpected short food availability (ZT13–14) was given at day 38, and expected food availability (ZT4–8) was omitted. Subsequently, 4h food availability was given at ZT 13–17.

Experiment II

Three male and 3 female Per1/2/3 triple KO mice (3–6 months old) were single-housed and maintained under a light-dark cycle with ad libitum access to food for 7 days. On day 8, mice were released to constant darkness. At day 12, mice were subjected to the restricted feeding procedure described above, except that restricted feeding was initially conducted every 21h (T21), then every 20h (T20), and then the phase of the feeding cycle was delayed for 6h.

Experiment III

Two male and 5 female Per1/2/3 triple KO mice (2.5–3.5 months old, male mice were heterozygous for cfos-shEGFP transgene) were single-housed and maintained under a light-dark cycle throughout the experiment. Mice initially had ad libitum access to food for 7 days, after which they were subjected to restricted feeding procedures described above (except onset of restricted feeding was set at ZT2). On 10 days of 4h restricted feeding, mice were given free access to food for 14 days, followed by 48h food deprivation.

Quantification and Statistical Analysis

Wheel-running activity, nose-poke, and pellet intake data were analyzed by ClockLab analysis software (version 6.1.18, Actimetrics, Lafayette, IN, USA). Double-plotted ethograms were generated in 6 min bins using the percentile plot with quantile 50. Group average 24-h profiles were generated with 6 min bins. As wheel-running activity and the rewarded left-nose-poke pattern are nearly identical, the left-nose-poke was used for quantitative analysis. Unless otherwise stated, qualitative statements are based on the left-nose-poke pattern. Experiment I: The period was estimated by regression line fitting to activity onsets (ClockLab). The onset of activity was determined in each day by default ClockLab setting (On hours: 5, Off hours: 5). The regression line was fitted during the days when the onset of activity aligned linearly. The accuracy of food anticipation was determined by the time when the first pellet intake was recorded during the daily scheduled meal. Experiment II: As very complex periodic components existed in the constant darkness, quantitative analysis was not possible. We classified food entertainment into 3 states: 1. Ultradian rhythm (no consolidated poking observed), 2. Unentrained (consolidated poking was not always observed prior to scheduled mealtime), 3. Entrained (consolidated poking always observed prior to scheduled mealtime). Experiment III: Group-averaged wheel-running, pellet intake, rewarded left-pokes, and unrewarded right-pokes are double plotted in 6 min bins with ClockLab percentile plot (quantiles = 50). To compare with current Per1/2/3 triple KO data, C57BL/6J wildtype data from our published paper (Ehichioya et al., 2025) was reanalyzed and plotted. Group-averaged ethograms of individual C57BL/6J mice from the first and second cohorts of our published data (Ehichioya et al., 2025) were generated in the same manner as for KO mice. As we intend to show deviation from the mean of each mouse, standard error of mean (S.E.M.) was used for the group-average profile. To account for individual variability, standard deviation (S.D.) was used to analyze first-pellet intake time during the scheduled meal. These are stated in the figure legend. As the current study is not a quantitative behavior study, statistical analysis was not conducted.

RESULTS

Experiment I

Consistent with our previous studies (Ehichioya et al., 2023; Pendergast et al., 2012), Per1/2/3 triple KO mice exhibit robust FAA and anticipatory left nose-poking behavior during restricted feeding under light-dark cycle. However, when the light cue was removed (constant darkness), FAA and anticipatory left nose-poking became less organized. Most of the Per1/2/3 triple KO mice (except #11664) did not consistently express continuous consolidated FAA and anticipatory left nose-poking behavior just before the feeding time (Fig. 1A, Fig. S1). Also consistent with our previous observations (Pendergast et al., 2012), mice exhibited rhythmic components with a period of ~22.3–23.7h in wheel-running and left nose-poking behaviors (Fig. 1A, Fig. S1 and 2). As anticipatory left nose-poking was present just prior scheduled mealtime, the first pellet intake during scheduled food availability was consistently observed at immediately after the onset of feeding time each day under light-dark cycle (Fig. 1B). However, when the light-dark cycle was removed, first pellet intake was observed each day at a variable time a few minutes after the onset of the scheduled meal (Fig. 1B). Somewhat surprisingly, mice without food anticipation were still able to eat pellets during restricted feeding: total daily pellet intake was no different under light-dark cycle and constant darkness (Fig. S3). Interestingly, in constant darkness, food-seeking rewarded left nose-poking progressively increased from shortly after the end of food availability until the time of food availability (Fig. S4). This contrasts with food-seeking behavior during restricted feeding under light-dark cycle, in which rewarded left nose-poking peaked before mealtime (Fig. S4). It is possible that Per1/2/3 triple KO mice use the FEO to predict mealtimes under light-dark cycle but use hunger-driven food-seeking (hourglass clock) to eat pellets under constant darkness (Per1/2/3 triple KO mice no longer expressed anticipatory food-seeking just prior to the onset of scheduled food availability in constant darkness). At the end of the experiment, mice were given unpredictable 1h food availability at ZT13, and subsequent predicted scheduled food availability (ZT4–8) was omitted. All mice took pellets during the short, unpredicted 1h of food availability only 5h after the previous mealtime, suggesting that mice were constantly seeking food even at an unpredicted time and without hunger. With the subsequent reward omission, despite the lack of anticipatory food-seeking behavior prior to the omitted mealtime, during the omitted scheduled mealtime at ZT4–8, mice poked both left and right nose-poke holes until unpredicted pellet access was given at ZT13–17. This observation cannot be explained by mice using a hunger-driven hourglass clock to take pellets during scheduled meals under constant darkness as, without reward, mice still exhibited increased food-seeking behavior that continued during predicted mealtime. Regardless, the current data is consistent with our previous observations that Per1/2/3 triple KO mice consistently exhibit consolidated anticipatory activity and food-seeking behavior just prior to daily recurring food availability under light-dark cycle. In constant darkness (absent of an environmental light cycle), however, these anticipations become disorganized and not always appear just prior to scheduled mealtimes under a 24h food availability cycle, likely because the 24h feeding cycle is outside of the range of food entrainment of the FEO in Per1/2/3 triple KO mice (Ehichioya et al., 2023; Pendergast et al., 2012).

Figure 1. Per1/2/3 triple KO mice failed to continue precise mealtime anticipation when the light-dark cycle was removed.

A: Double-plotted ethograms (rewarded left pokes) of all 6 individual Per1/2/3 triple KO mice during 24h restricted feeding under light-dark cycle and subsequent constant darkness. The light-dark cycle is indicated with white and black bars at the top of each ethogram. The day when mice were released to constant darkness is indicated on the left side of each ethogram. Food availability is shown with a dark gray outlined box, on the left half of the double-plotted ethograms (food availability of the last 2 days is also shown on the right half of the ethograms). Ethograms (wheel running, pellet intake, rewarded left poke, and unrewarded right poke) of all individual mice during the entire experiment are shown in Figure S1. B: Mean ± S.D. of the time when first pellet intake occurred during each scheduled feeding cycle is shown. Scheduled daily meal started at Time 4.0. LD: 12h light and 12h dark. DD: constant darkness.

Experiment II

To measure the range of food entrainment of the FEO in Per1/2/3 triple KO mice, we next conducted experiments in which Per1/2/3 triple KO mice were subjected to a shorter feeding cycle under constant darkness (Fig. 2, Fig. S5, and Table 1). KO mice initially had ad libitum access to food under light-dark conditions, then they were released to constant darkness. All mice exhibit arrhythmicity or ultradian rhythm in left nose-pokes under constant darkness. Mice were then given first a T21 feeding cycle and subsequently a T20 feeding cycle, after which the T20 feeding cycle was phase-delayed by 6h. During the T21 feeding cycle, mouse #11591 exhibited consolidated food-seeking activity with a period deviated from the 21h feeding cycle. Under the T20 feeding cycle and subsequent phase-delayed T20 feeding cycle, the left nose-poking rhythm didn’t entrain to the 20h feeding cycle. Mouse #11592 initially expressed consolidated and entrained anticipatory left nose-poke food-seeking behavior under T21 feeding, but after ~5 cycles, the consolidated food-anticipatory feed-seeking disappeared. Upon switching to T20 feeding, the mouse re-expressed robust entrained food-seeking behavior. This rhythm re-entrained to the delayed feeding cycle. Mouse #11593 exhibited entrained food-seeking behavior rhythm under both the T21 and T20 feeding cycles. However, upon delaying the phase of the T20 feeding cycle, anticipatory food seeking appeared to advance in phase but wasn’t able to re-entrain to the shifted T20 cycle. Mouse #11684 exhibited a rhythm period deviated from the T21 feeding cycle. The rhythm entrained to the T20 feeding cycle and re-entrained to the shifted T20 cycle. Mice #11687 and #11898 exhibited rhythms with periods deviated from the feeding cycle throughout the experiment, and no entrainment was observed. For all mice, the patterns of wheel-running and unrewarded right poking behaviors were generally similar to those of rewarded left poking behavior (Fig. S4). Overall, Per1/2/3 KO mice occasionally exhibited anticipatory activity rhythms entrained to the feeding cycle, suggesting the FEO is food-entrainable in the absence of a light-dark cycle.

Figure 2. Double-plotted ethograms (rewarded left pokes) of all 6 individual Per1/2/3 triple KO mice during T-feeding cycles under constant darkness.

The same data is plotted with two different time bases (Top: modulo 21h, Bottom: modulo 20h). Food availability is indicated as a dark gray outlined box on the left half of the double-plotted ethograms. Ethograms (wheel running, pellet intake, rewarded left poke, and unrewarded right poke) of all individual mice during the entire experiment are shown in Figure S5.

Table 1.

Summary of the rhythmic properties of Per1/2/3 triple KO mice under multiple T-feeding cycles.

Animal ID	Age	Sex	Strain	Feeding cycle
				T21	T20	T20 delay
11591	24 weeks old	F	C57BL/6J	+	+	+
11592	24 weeks old	F	C57BL/6J	++ → •	++	++
11593	24 weeks old	F	C57BL/6J	++	++	+
11684	18 weeks old	M	C57BL/6J	+	++	++
11687	18 weeks old	M	C57BL/6J	+	+	+
11898	12 weeks old	M	C57BL/6J	+	+	+

• No consolidation (ultradian rhythm)

+ Consolidated (unentrained)

++ Consolidated (entrained)

Experiment III

Using C57BL/6J wildtype mice, we previously hypothesized that food-seeking anticipatory behavior and food intake are controlled by different circadian oscillators (Fig. S6) (Ehichioya et al., 2025). Anticipatory food-seeking behavior is controlled by the light entrainable extra-SCN circadian oscillator (Oscillator II, aka FEO). Feeding is controlled by another oscillator, which is coupled with the SCN (Oscillator I), as daytime restricted feeding–entrained pellet intake during the day gradually shifted back to night during the subsequent period of ad libitum feeding. We previously showed that Oscillator II is functional in Per1/2/3 triple KO mice (Ehichioya et al., 2023). Here, we test if Oscillator I is functional in Per1/2/3 triple KO mice. Per1/2/3 triple KO mice were subjected to restricted feeding under light-dark cycle with the same procedures we used for wildtype mice (Fig. S6) (Ehichioya et al., 2023), except we set restricted feeding from ZT2–6, as activity onset of Per1/2/3 triple KO mice usually starts at ~2h before lights-off (Fig. 3, Fig. S7). During ad libitum feeding, Per1/2/3 triple KO mice exhibited wheel-running activity, pellet intake, rewarded and unrewarded nose-pokes ~2h before lights-off. Interestingly, we observed masking effects of lights-off. At the time of lights-off, wheel-running activity transiently increased while pellet intake and rewarded left nose-pokes were transiently suppressed (Fig. 3, Fig. S8). As in wildtype mice, Per1/2/3 triple KO mice exhibited robust food-anticipatory activity and poking during daytime restricted feeding. In contrast with wildtype, in which pellet intake gradually moves back to night in subsequent ad libitum feeding (Fig. S6B), Per1/2/3 triple KO mice continue to eat during day and night (Fig. 3B). Per1/2/3 triple KO mice appeared to continue to express anticipatory activity and poking at the previous feeding time (Fig. S8). Robust anticipatory activity and poking were observed during subsequent food deprivation (Fig. 3, Fig. S9). This suggests Oscillator II (aka FEO) is intact in Per1/2/3 triple KO mice. As Per1/2/3 triple KO mice initially exhibit nocturnal pellet intake, Oscillator I in Per1/2/3 triple KO mice is likely to be functional. In contrast to in wildtype, phase-advanced pellet intake during restricted feeding didn’t return to nighttime in Per1/2/3 triple KO mice. We interpreted this result to be a result of the SCN in Per1/2/3 triple KO mice is not rhythmic under light-dark cycle (Ehichioya et al., 2023), the SCN was unable to pull back the phase of Oscillator I at night upon release to ad libitum feeding.

Figure 3. Group-averaged double-plotted ethograms of Per1/2/3 triple KO mice with 24h feeding and light-dark cycles.

Group averaged wheel running, pellet intake, rewarded left poke, and unrewarded right poke are double plotted in 6 min bin with ClockLab percentile plot (quantiles = 50). The light-dark cycle is indicated with white and black bars at the top of each ethogram, and time of food availability is shown dark gray outlined box on the left side of each double-plotted ethogram. Ethograms of individual mice are shown in Figure S7.

DISCUSSION

Food is a weak Zeitgeber for the FEO.

Our previous and current studies showed that Per1/2/3 triple KO mice were able to anticipate 24h feeding cycle under 12h light: 12h dark conditions. However, when the environmental light cycle was removed by releasing the mice to constant darkness, Per1/2/3 triple KO mice failed to exhibit consolidated FAA and anticipatory nose-poking behavior just prior to the scheduled mealtime. Mice still exhibited consolidated activity and nose-poking, but these were disorganized and didn’t always align with just before scheduled food availability. This is in contrast with a similar study with wildtype mice, in which wildtype mice continued to express FAA just before scheduled mealtime even after being released to constant darkness (Castillo et al., 2004). The difference between the two studies can be explained by the difference in the autonomous period between Per1/2/3 KO and wildtype mice (Pendergast et al., 2012). The regularity of the first pellet intake during scheduled food availability was also altered by the removal of light-dark cycle. Under light-dark cycle, first pellet intake was consistently observed immediately after onset of food availability during light-dark cycle, but under constant darkness, it became inconsistent and often delayed. Therefore, the presence of anticipatory food-seeking prior to predicted food availability is essential to taking the first reward as soon as the scheduled mealtime starts. However, it is surprising that mice still managed to take the essential daily food during 4h of scheduled feeding time without showing consolidated anticipatory nose-poking behavior immediately prior to mealtime. With the current FED3 settings, mice have to poke the left nose-poke hole during food availability to dispense a pellet. As anticipatory food-seeking was not always present before scheduled mealtime in constant darkness, a hunger-driven hourglass clock and/or opportunistic food-seeking might be involved in obtaining food pellets during scheduled food availability. However, during reward omission, though anticipatory nose-poking was absent prior to the omitted mealtime, during the omitted feeding time, mice exhibited intense nose-poking which continued until the mice received a food reward, which suggests that mice were still predicting the omitted mealtime, even without exhibiting consolidated anticipatory nose-poking immediately before it. There must another time-keeping mechanism or time memory involvement in omitted predicted food availability.

Our current T-feeding cycle under constant darkness study replicated Fred Stephan’s historical experiments that showed the FAA of SCN-lesioned rats entrained to feeding cycle (Stephan, 1992; Stephan et al., 1979a; Stephan et al., 1979b); additionally, we were also able to replicate Stefan’s result that food entrainment can only be observed in a very narrow range of feeding cycle periods close to the autonomous period of the FEO. When the light-dark cycle is present, Per1/2/3 triple KO mice were able to anticipate a 24h feeding cycle. As retinal projections to the brain are not disrupted in Per1/2/3 triple KO mice, KO mice do exhibit daily rhythms under a light-dark cycle. Our previous and current studies further suggest a potentially important property of the FEO, that the primary environmental signal to which the FEO entrains is light, and feeding time is similarly encoded to the FEO as it encodes time memory in honeybees (Beer et al., 2024; Beling, 1929; Renner, 1955; Renner, 1957; Renner, 1960; Wahl, 1932). As the FEO can entrain to a 24h light-dark cycle, the range of light entrainment of the FEO is much wider than the range of food entrainment is.

Ecological significance of the use of the environmental light-dark cycle to anticipate food availability.

We propose our hypothesis that organisms use the most reliable environmental daily information, “light”, to anticipate the time of food availability. Because the time of food availability can often change over time in nature, entraining the phase of the pacemaker to the time of food availability is not the best strategy to predict food availability. Instead, by encoding mealtime engrams to a light-entrainable oscillator, organisms are able to anticipate the ever-changing time of food availability quickly and flexibly. Although anticipation of multiple mealtimes per day can be achieved by multiple oscillators, each of which controls the anticipation of each mealtime (Petersen et al., 2022), our proposed hypothesis provides an alternative model that a single pacemaker can anticipate multiple times of food resource availability at different times of the day.

A novel strategy to identify candidate brain loci for the FEO.

The location of the FEO has been a longstanding mystery for many decades. Over the past 50 years, researchers have sought to identify the location of the pacemaker in the brain and even in peripheral organs. However, such attempts were unsuccessful (Davidson, 2009; Davidson et al., 2003; Mistlberger, 1994; Stephan, 2002). The name “food-entrainable oscillator” has led researchers into searching for an oscillator that entrains to the time of food consumption. Although there have been many food-entrainable oscillators identified in the brain and in peripheral tissues, none of these oscillators have shown to be the FEO. Our model provides a new, directly feasible approach to the search for the oscillator—instead of searching for an oscillator that entrains to food, searching novel brain locations activated by light or oscillators that entrains to light, possibly will lead to fruitful outcomes. In the past, researchers have used the light input to discover the neuroanatomical location of the circadian pacemakers (Menaker et al., 1978; Moore and Eichler, 1972; Saunders, 1976; Stephan and Zucker, 1972). As the FEO primarily uses light to entrain the pacemaker to the environment, the same strategy can be used to identify candidate loci of the FEO.