Wistar rats and C57BL/6 mice differ in their motivation to seek social interaction versus food in the Social versus Food Preference Test

Abstract

Here we characterized the Social versus Food Preference Test, a behavioral paradigm designed to investigate the competition between the choice to seek social interaction versus the choice to seek food. We assessed how this competition was modulated by internal cues (social isolation, food deprivation), external cues (stimulus salience), sex (males, females), age (adolescents, adults), and rodent model (Wistar rats, C57BL/6 mice). We found that changes in stimulus preference in response to the internal and external cue manipulations were similar across cohorts. Specifically, social over food preference scores were reduced by food deprivation and social familiarly in Wistar rats and C57BL/6 mice of both sexes. Interestingly, the degree of food deprivation-induced changes in stimulus investigation patterns were greater in adolescents compared to adults in Wistar rats and C57BL/6 mice. Strikingly, baseline stimulus preference and investigation times varied greatly between rodent models: across manipulations, Wistar rats were generally more social-preferring and C57BL/6 mice were generally more food-preferring. Adolescent Wistar rats spent more time investigating the social and food stimuli than adult Wistar rats, while adolescent and adult C57BL/6 mice investigated the stimuli a similar amount. Social isolation did not alter behavior in the Social versus Food Preference Test. Together, our results indicate that the Social versus Food Preference Test is a flexible behavioral paradigm suitable for future interrogations of the peripheral and central systems that can coordinate the expression of stimulus preference related to multiple motivated behaviors.

GRAPHICAL ABSTRACT

1 |. INTRODUCTION

Behavior is influenced by a combination of internal and external cues, with the expression of the appropriate behavior dependent on an individual’s current motivational state and the presence of stimuli in their surrounding environment [1, 2]. Thus far, most laboratory studies have focused on uncovering the peripheral and central systems that regulate the expression of a single behavior or the expression of a suite of behaviors associated with a single motivational state. In natural settings, however, an individual can be simultaneously experiencing multiple motivational states with multiple choices of how to act. Yet, the direct assessment of the roles of peripheral and central systems in coordinating motivated behavioral choice is largely understudied [3–8]. This may be due to a lack of behavioral tests that are suitable for such investigations. Here, we characterized a recently developed behavioral paradigm [3, 9], hereafter called the Social versus Food Preference Test, to test the competition between the choice to seek social interaction versus the choice to seek food.

To determine how internal cues modulate the competition between the choice to seek social interaction versus the choice to seek food, we altered the motivational states of subjects by exposing them to acute social isolation and/or acute food deprivation prior to exposure to the Social versus Food Preference Test. To examine whether the effects of these manipulations were similar between the sexes, stable across the lifespan, and comparable between commonly used laboratory rodent models, experiments were conducted with adolescent and adult Wistar rats and C57BL/6 mice of both sexes (Experiments 1 and 2). For all subjects, we predicted that social isolation biases preference towards the social stimulus, and that food deprivation biases preference towards the food stimulus.

To determine how external cues modulate the competition between the choice to seek social interaction versus the choice to seek food, we tested whether the saliency of the social stimulus alters social versus food preference in adolescent Wistar rats and C57BL/6 mice of both sexes (Experiment 3). We predicted that stimulus preference is more biased toward the social stimulus when the social stimulus is novel compared to when the social stimulus is familiar [10–12].

Lastly, to validate general sociability in C57BL/6 mice [11], we tested the preference of adolescent and adult mice of both sexes to investigate a social stimulus versus an empty corral (Experiment 4). In sum, the series of experiments presented in this paper aimed to characterize a flexible behavioral paradigm suitable for future interrogations of the peripheral and central systems that coordinate the choice to seek social interaction versus the choice to seek food.

2 |. MATERIALS AND METHODS

2.1 |. Animals

Upon arrival to the animal care facility at Michigan State University, male and female Wistarrats (Charles River Laboratories) were housed in single sex pairs in standard rat cages (48 × 27 × 20 cm), and male and female C57BL/6 mice (The Jackson Laboratory’s stock 000664 or Charles River Laboratories) were housed in single sex groups of two or four in standard mouse cages (29 × 19 × 13 cm). Rats and mice were housed in separate colony rooms within the vivarium, and all animals were maintained under standard laboratory conditions (12 hr light/dark cycle; water ad libitum; food ad libitum except as described in Section 2.3). All housing and testing was in accordance with the National Institute of Health Guidelines for C are and Use of Laboratory Animals and the Michigan State University Institutional Animal Care and Use Committee (IACUC). Animals were acclimated to the colony rooms for at least 72 hrs prior to the start of daily handling or any experimental procedures. All animals were gonadally intact, but estrous cycle was not monitored.

2.2 |. Social versus Food Preference Test

The Social versus Food Preference Test was used to assess the preference of Wistar rats and C57BL/6 mice to investigate a social stimulus (age-, sex-, and species-matched conspecific) versus a food stimulus (standard laboratory chow; Teklad Irradiated 22/5 Rodent Diet, 8940) [9]. This test was based on our previously developed social novelty preference test in rats [10], and a social interaction assay used to determine the effects of hunger signals on social interest in mice [3]. Social versus food preference was tested using a 3-chambered apparatus, where the social stimulus and the food stimulus were located on opposite ends. Two sizes of this apparatus were custom-constructed (Fig 1), one sized for rats (Scientific Instrumental and Machining Services, Boston College) and one sized for mice (Physics and Astronomy Machine Shop, Michigan State University), and each testing apparatus was located in its respective colony room within the vivarium. The exterior of the apparatus was composed of acrylic (rats) or PVC (mice), and each chamber (rats: 40 cm × 40 cm × 27 cm; mice: 30 cm × 30 cm × 20 20 cm) was separated by a translucent acrylic partition with an opening (rats: 10 cm × 10.2 cm; mice: 5 cm × 5 cm) to allow passage between chambers. Stimuli were placed in corrals, which allowed for olfactory, visual, and auditory contact, but restricted tactile contact of the stimuli by the experimental subject. Minimal nose or forepaw contact between the experimental and stimulus animals was possible and the experimental animals sometimes sniffed the tail of the stimulus animal if it extended outside of the corral, but no anogenital investigation, playful behaviors, or aggressive behaviors were observed. Food pellets were moved away from accessible edges to prevent consumption. For rats, rectangular corrals (18 cm × W 10 cm D × 21 H cm) were composed of a solid translucent acrylic top/bottom/back and translucent acrylic bars (0.6 cm diameter, spaced 1.75 cm apart center-to-center) on the other three sides. For mice, cylindrical corrals (8.5 cm ID, 10.5 cm OD × 17cm H) were composed of solid translucent acrylic top/bottom connected by translucent acrylic bars (0.6 cm diameter, spaced 1.5 cm apart center-to-center). The apparatus was cleaned with 70% ethanol and corrals were cleaned with dilute cleaning solution at the start and end of each day, as well as between subjects.

Fig 1. Social versus Food Preference Test.

Rats (top) and mice (bottom) were placed into the center of a 3-chambered apparatus and then allowed to freely investigate a social stimulus and a food stimulus, which were placed in corrals located on opposite ends, for a period of 10 min.

All subjects were habituated to the testing procedures the day prior to their first test. During habituation, experimental animals were placed into the center chamber and allowed to freely explore the apparatus and investigate empty corrals for 10 min before being returned to their homecage. In separate trials, stimulus animals were habituated to confinement within a corral for 10 min before being returned to their homecage. There were no significant chamber preferences during habituation in any experiment (data not shown), and data from habituation sessions were not used in any other statistical analyses.

During testing, the experimental subject was placed into the center chamber and allowed to freely explore the apparatus and investigate the corralled stimuli for 10 min. To reduce the number of animals used, when the social stimulus was novel it was used twice per day (exposed to two different experimental subjects in non-successive tests); food pellets were replaced between subjects. The location of the social and food stimuli (i.e., left chamber or right chamber) was pseudorandom and counterbalanced between subjects each day and within subjects across test days (when applicable). A webcam (Logitech HD Pro C910) was attached to the ceiling and connected to a PC computer in an adjoining room to record the sessions. Experimenters, who were unaware of sex and testing conditions, scored recorded videos using Solomon Coder (https://solomon.andraspeter.com/) to quantify the amount of time the experimental subjects spent investigating each of the two stimuli, and these values were then summed to calculate total stimulus investigation time. Investigation was defined as when the subject was actively engaged with the corral (e.g., sticking nose between bars, pawing, sniffing) with its attention directed towards the stimulus inside of the corral as indicated by head position/gaze orientation. To obtain a measure of stimulus preference, the percent of time the subject investigated the social stimulus [(time spent investigating social stimulus/total stimulus investigation time)*100] was calculated (“social over food preference score”). Values > 50% indicate that subjects spent more time investigating the social stimulus, and values < 50% indicate that subjects spent more time investigating the food stimulus. AnyMaze (Stoelting) was used to quantify measures of locomotor activity (i.e., middle chamber entries, distance traveled). In instances where AnyMaze failed to track the experimental subject, experimenters manually scored videos for middle chamber entries and these subjects were removed from the distance traveled analyses (Experiment 1a: n = 5 females; Experiment 1b: n = 1 male, n = 4 females, Experiment 2a: n = 2 males, n = 2 females, Experiment 2b: n = 1 female, Experiment 3b: n = 1 female).

2.3 |. Experimental Procedures

2.3.1 |. Experiment 1: The effects of social isolation and food deprivation on social versus food preference in rats

To establish the Social versus Food Preference Test and characterize how manipulating the internal motivational drives for social interaction-seeking and food-seeking behavior affects behavior in this test, we tested how subjects’ preference to investigate a social stimulus (novel age-, sex-, and species-matched conspecific that was maintained under pair-housing conditions) versus a food stimulus (standard laboratory chow) was modulated by social isolation and hunger. In Experiment 1a the subjects were adolescent (39–46 day old) Wistar rats (8 males/6 females; 1 male was subsequently removed from all analyses due to escaping the testing apparatus), and in Experiment 1b the subjects were a separate cohort of adult (13–14 week old) Wistar rats (8 males/8 females). In both experiments, subjects were first habituated to the testing apparatus as described above, and then tested in the Social versus food Preference Test at zeitgeber time (ZT) 12 on four occasions each 48 hrs apart using a within-subjects 2 × 2 counterbalanced design (pair-housed/socially isolated × sated/food-deprived). The length of social isolation and/or food deprivation was 24 hrs, and experimental subjects were exposed to a different unfamiliar social stimulus during each test. Subjects’ body weights during this 24 hr period were monitored as a proxy physiological measure of hunger. Subjects were returned to their homecage immediately following the Social versus Food Preference Test, and ad lib chow consumption during the subsequent 30 min was monitored to behaviorally assess hunger [9]. As such, when subjects were tested under pair-housed conditions consumption was measured as the amount eaten by the pair, and when subjects were tested under socially isolated conditions consumption was the amount eaten by each individual. Percent change in body weight [((end weight-start weight)/start weight)*100] and consumption as a percent of body weight [(total grams consumed/total grams body weight)*100] were calculated to normalize the data across experiments and account for baseline sex differences in body weight which could also influence consumption potentially due to differences in gastric capacity [13] or other factors [14]. Lastly, food was removed from the stimulus animals’ homecages 2 hrs prior to the start of the testing each day to reduce the amount of food-related sensory cues present on the social stimuli [as described in: 3]. At the end of each test session, animals were returned to their original pair-housing conditions with ad lib food access.

2.3.2 |. Experiment 2: The effects of social isolation and food deprivation on social versus food preference in mice

To determine whether the stimulus preference and investigation patterns observed in Experiment 1 are conserved across commonly used laboratory rodent models, we repeated this experiment in C57BL/6 mice. In Experiment 2a the subjects were adolescent (37–43 day old) C57BL/6 mice (8 males/8 females; 1 female was subsequently removed from all analyses for spending, on average, 3 min of each test behind a corral and out-of-view of the experimenter), and in Experiment 2b the subjects were a separate cohort of adult (13–14 week old) C57BL/6 mice (8 males/8 females; 1 female was subsequently removed from all analyses for failure to explore all three chambers during habituation or test one). Experimental subjects were switched from group-housing (4 per cage) to pair-housing conditions following habituation, while stimulus animals were maintained 4 per cage. The following changes were implemented: adolescent subjects were tested 2 days younger than in Experiment 1a to account for the mildly faster development of mice compared to rats [15], testing started at ZT7 (based on outcomes of [9]), the length of social isolation/food deprivation was reduced to 18 hrs per IACUC recommendation and following pilot testing to identify conditions that ensured mice would not lose more than 15% of their body weight, and post-test food consumption was conducted individually in clean cages to allow for within-subjects comparison of consumption by subjects tested under pair-housed and socially isolated conditions (after the consumption test, mice were re-housed in their homecages under their original pair-housing conditions).

2.3.3 |. Experiment 3: Modulation of social versus food preference by social salience in sated adolescent rats and mice

To determine whether stimulus preference and investigation patterns is modulated by the salience of external stimuli, we tested subjects’ preferences to investigate a novel (age-, sex-, and species-matched conspecific that was maintained under pair-housing conditions) or familiar (cagemate) social stimulus versus the food stimulus (standard laboratory chow) in experimental subjects that were also maintained under pair-housing and ad lib feeding conditions. Subjects were first habituated to the testing apparatus and then using a within-subjects counterbalanced design each subject was tested on two occasions 48 hrs apart, with testing starting at ZT7. Since adolescents showed greater changes in behavior compared to adults in Experiments 1 and 2, only adolescent subjects were used in Experiment 3. In Experiment 3a, the subjects were adolescent (39–42 day old) Wistar rats (9 males/8 females), and in Experiment 3b the subjects were adolescent (37–40 day old) C57BL/6 mice (8 males/8 females). Cagemates were not used as novel social stimuli for other experimental subjects. Because subjects were tested under sated conditions and cagemates sometimes served as stimulus animals, food was not removed from the homecages of stimulus animals prior to testing.

2.3.4 |. Experiment 4: Do sated mice prefer to investigate a novel social stimulus over an empty corral?

In contrast to Wistar rats, C57BL/6 mice did not exhibit a social preference in Experiments 2 or 3. Thus, to validate general sociability in C57BL/6 mice, we tested subjects’ preferences to investigate a social stimulus (novel age-, sex-, and species-matched conspecific) versus an empty corral. Subject mice were maintained under pair-housing and ad lib feeding conditions, habituated to the testing apparatus, and tested the following day starting at ZT7. In Experiment 4a the subjects were adolescent (45 day old) C57BL/6 mice (4 male/4 female; formerly stimulus animals in Experiment 2a), and in Experiment 4b the subjects were a separate cohort of adult (15 week old) C57BL/6 mice (4 male/4 female; formerly stimulus animals in Experiment 2b). In both experiments, experimental subjects and stimulus animals were placed in pair-housing conditions at least 48 hrs prior to testing. As in Experiments 1 and 2, food was removed from the homecages of the stimulus animals 2 hrs prior to the start of testing.

2.4 |. Statistical analysis

Mixed-model [sex (male, female; between-subjects factor) × hunger condition (food-deprived, sated; within-subjects factor) × housing condition (socially isolated, pair-housed; within-subjects factor)] ANOVAs were used to assess the effects of sex, food deprivation, and social isolation on stimulus preference, stimulus investigation, locomotor activity, and body weight measures in Experiments 1a, 1b, 2a, and 2b, as well as consumption measures in Experiments 2a and 2b. Since consumption was assessed under pair-housed conditions in Experiments 1a and 1b (see Section 2.3.1), separate mixed-model [sex (male, female; between-subjects factor) × hunger condition (food-deprived, sated; within-subjects factor)] ANOVAs were used to evaluate consumption under social isolation and pair-housed conditions. Mixed-model [sex (male, females; between-subjects factor) × social saliency (novel, cagemate; within-subjects factor)] ANOV As were used to assess the effects of sex and social salience for all measures in Experiments 3a and 3b. Mixed-model ANOVAs [sex (male, female; between-subjects factor) × stimulus (social corral, empty corral; within-subjects factor)] investigated sex differences in corral investigation times in Experiments 4a and 4b. Independent samples t-Tests were used to assess sex differences in stimulus preference and locomotor activity in Experiments 4a and 4b. When significant interactions were found in the ANOVAs, Bonferroni post hoc pairwise comparisons were conducted to clarify the effects. For all experiments, one-sample t-Tests with a reference value of 50% were used to evaluate stimulus preference. Normality was assessed by Shapiro-Wilk testing of the standardized residuals, sphericity by Mauchly’s Test, and equality of variances by Levene’s Test. All data were analyzed using IBM SPSS Statistics 24–26, and statistical significance was set at p < 0.05. Cohen’s d (d) effect sizes were manually computed for all t-Tests, and partial eta squared (η²) effect sizes were computed in SPSS for all ANOVAs. Graphs were produced in Microsoft Excel using custom templates and those from [16], and then edited in Adobe Illustrator CC.

3 |. RESULTS

3.1 |. Experiment 1a: Social versus food preference was altered by food deprivation, but not social isolation in adolescent rats

Adolescent rats had significantly lower social over food preference scores when tested under food-deprived compared to sated conditions; neither housing condition nor sex altered stimulus preference (Table 1, Fig 2A, B). Adolescent rats had a strong social preference under sated conditions (pair-housed: t₍₁₂₎ = 20.0, p < 0.001, d = 5.55; socially isolated: t₍₁₂₎ = 17.1, p < 0.001, d = 4.74), and no stimulus preference under food-deprived conditions (pair-housed: t₍₁₂₎ = 1.40, p = 0.19, d = 0.39; socially isolated: t₍₁₂₎ = 0.14, p = 0.89, d = 0.039; Fig 2A, B).

Table 1.

ANOVA statistics and partial eta squared (η²) effect sizes for Experiment 1a: Social versus food preference was altered by food deprivation, but not social isolation in adolescent rats. Significant effects shown in bold, n.s. = none significant, n.a. = not applicable, BW = body weight.

	Sex	Hunger Condition	Housing Condition	Interactions
Social over Food Preference [%]	F(1,11) = 1.95, p = 0.19, η2 = 0.15	F(1,11) = 58.5, p <0.001, η2 = 0.84	F(1,11) = 1.36, p = 0.27, η2 = 0.11	n.s.
Social Stimulus Investigation [sec]	F(1,11) = 0.84, p = 0.38, η2 = 0.071	F(1,11) = 9.83, p = 0.009, n2 = 0.47	F(1,11) = 0.00, p = 0.99, η2 < 0.001	n.s
Food Stimulus Investigation [sec]	F(1,11) = 2.77, p = 0.12, η2 = 0.20	F(1,11) = 88.6, p < 0.001, η2 = 0.89	F(1,11) = 3.05, p = 0.11, η2 = 0.22	n.s.
Total Investigation [sec]	F(1,11) = 0.001, p = 0.98, η2 = 0.001	F(1,11) = 1.18, p = 0.30, η2 = 0.097	F(1,11) = 1.23, p = 0.29, η2 = 0.10	n.s.
Middle Chamber Entries [#]	F(1,11) = 1.97, p = 0.19, η2 = 0.15	F(1,11) = 15.5, p = 0.002, η2 = 0.59	F(1,11) = 1.92, p = 0.19, η2 = 0.15	n.s.
Body Weight [% change]	F(1,11) = 2.03, p = 0.18, η2 = 0.16	F(1,11) = 241,p < 0.001, η2 = 0.96	F(1,11) = 11.7, p = 0.006, η2 = 0.51	n.s.
Consumption [% of BW] – Pair-Housed	F(1,11) = 0.53, p = 0.51, η2 = 0.12	F(1,11) = 111, p < 0.001, η2 = 0.97	n.a.	n.s
Consumption [% of BW] – Socially Isolated	F(1,11) = 1.94, p = 0.19, η2 = 0.15	F(1,11) = 31.3, p < 0.001, η2 = 0.74	n.a.	n.s

Fig 2. Experiment 1a.

Food deprivation abolished preference for the social stimulus in adolescent rats (A), while social isolation had no effect on stimulus preference (B). Food deprivation decreased investigation of the social stimulus, and robustly increased investigation of the food stimulus (C). Representative heat maps of activity from one subject (D). Bar graphs (mean ± SEM) collapsed across sex for A and B (same data replotted to illustrate main effects), and across sex and housing condition for C; individual data collapsed across housing condition in C; * p < 0.05, ** p < 0.01, *** p < 0.001 mixed-model ANOVA; ^ p < 0.05, one-sample t-Test from 50% (gray dashed line); n.s. = not significant; n = 7 males, n = 6 females.

The observed changes in stimulus preference were due to decreased time spent investigating the social stimulus and increased time spent investigating the food stimulus when adolescent rats were food-deprived compared to when they were sated (Table 1, Fig 2C). Despite food deprivation-induced changes in the time spent investigating each stimulus, there was no net change in total (social+ food) investigation time as a result of food deprivation (Table 1, Fig 2C). Stimulus investigation times were unaffected by housing condition or sex (Table 1).

Locomotor activity, as measured by middle chamber entries, was decreased under food-deprived compared to sated conditions and unaffected by housing condition or sex in adolescent rats (Tables 1, 2). Distance traveled was not assessed due to technical difficulties tracking the female cohort (see Section 2.2).

Table 2.

Activity measures, body weight, and post-test food consumption results for Experiments 1 and 2. Data shown as mean ± SEM. Distance traveled was not assessed in Experiment 1a due to technical difficulties. ¹main effect of sex, ²main effect of hunger condition, ³main effect of housing condition, ⁴significant sex × hunger interaction, ⁵significant hunger × housing interaction, ⁶significant sex × hunger × housing interaction; see text for details and Tables 1, 3–5 for corresponding ANOVA statistics.

		Sated & Pair-Housed	Sated & Socially Isolated	Food-Deprived & Pair-Housed	Food-Deprived & Socially Isolated
Total Distance Traveled [m]
Exp 1b: Adult Rats	Males	33.2 ± 1.87	30.1 ± 2.72	31.3 ± 2.72	33.2 ± 1.33
	Females	36.4 ± 4.12	40.0 ± 4.41	35.5 ± 2.20	40.6 ± 3.70
Exp 2a: Adolescent Mice3	Males	22.4 ± 1.09	21.5 ± 1.47	20.5 ± 1.30	20.7 ± 1.35
	Females	23.8 ± 2.34	23.5 ± 2.66	26.1 ± 1.87	19.6 ± 1.87
Exp 2b: Adult Mice1,2	Males	16.7 ± 0.99	17.8 ± 1.57	13.4 ± 1.41	18.3 ± 2.24
	Females	23.0 ± 2.44	23.3 ± 2.06	20.3 ± 1.85	19.2 ± 2.16
Middle Chamber Entries [#]
Exp 1a: Adolescent Rats2	Males	37.0 ± 2.79	35.7 ± 3.17	30.7 ± 2.73	29.0 ± 1.96
	Females	34.2 ± 2.54	31.7 ± 2.65	27.3 ± 2.23	24.3 ± 3.87
Exp 1b: Adult Rats4	Males	32.4 ± 1.81	27.6 ± 3.12	26.9 ± 2.16	31.8 ± 2.08
	Females	27.5 ± 1.45	31.6 ± 2.40	27.5 ± 2.28	29.8 ± 1.60
Exp 2a: Adolescent Mice	Males	27.5 ± 2.19	27.6 ± 2.20	25.8 ± 0.75	25.5 ± 1.31
	Females	30.0 ± 3.43	28.7 ± 3.39	26.6 ± 3.04	22.4 ± 2.32
Exp 2b: Adult Mice1,2,4	Males	15.9 ± 1.32	17.9 ± 1.69	13.0 ± 2.13	16.8 ± 2.76
	Females	28.1 ± 3.42	29.6 ± 4.11	25.3 ± 3.23	19.7 ± 1.94
Body Weight [% Change]
Exp 1a: Adolescent Rats2,3	Males	4.58 ± 0.56	3.36 ± 0.90	−9.42 ± 0.52	−11.20 ± 0.96
	Females	4.36 ± 1.23	3.28 ± 0.61	−10.17 ± 0.80	−12.97 ± 1.21
Exp 1b: Adult Rats1,2	Males	0.73 ± 0.39	0.51 ± 0.33	−5.66 ± 0.25	−5.74 ± 0.28
	Females	0.34 ± 0.31	0.70 ± 0.42	−7.26 ± 1.10	−8.02 ± 0.83
Exp 2a: Adolescent Mice2,3	Males	2.30 ± 0.84	2.57 ± 0.54	−10.71 ± 0.79	−14.32 ± 1.31
	Females	2.91 ± 0.81	2.47 ± 1.23	−11.56 ± 0.99	−13.52 ± 1.17
Exp 2b: Adult Mice2,5	Males	−1.19 ± 0.56	2.41 ± 0.69	−9.19 ± 0.58	−13.41 ± 1.63
	Females	−0.43 ± 1.02	1.16 ± 0.19	−9.84 ± 1.30	−12.50 ± 1.51
Consumption [% of Body Weight]
Exp 1a: Adolescent Rats2	Males	0.62 ± 0.35	1.14 ± 0.46	3.34 ± 0.31	3.04 ± 0.57
	Females	0.68 ± 0.15	0.64 ± 0.26	3.65 ± 0.17	3.33 ± 0.29
Exp 1b: Adult Rats2	Males	0.44 ± 0.10	0.61 ± 0.23	1.53 ± 0.18	1.25 ± 1.12
	Females	0.56 ± 0.13	0.40 ± 0.12	1.88 ± 0.32	1.72 ± 0.09
Exp 2a: Adolescent Mice1,2,4	Males	0.38 ± 0.07	0.36 ± 0.09	2.33 ± 0.17	2.41 ± 0.17
	Females	0.16 ± 0.06	0.40 ± 0.15	2.70 ± 0.36	3.36 ± 0.32
Exp 2b: Adult Mice2	Males	0.66 ± 0.11	0.59 ± 0.11	2.41 ± 0.17	2.35 ± 0.34
	Females	0.72 ± 0.15	0.59 ± 0.14	2.67 ± 0.37	2.60 ± 0.27

Adolescent rats lost weight under food deprivation and gained weight under sated conditions, resulting in a significant main effect of hunger condition on percent change in body weight (Tables 1, 2). There was also a significant main effect of housing condition on percent change in body weight; under socially isolated conditions adolescent rats lost more weight when food-deprived and gained less weight when sated compared to under pair-housed conditions (Tables 1, 2). Adolescent rats consumed significantly more food when food-deprived compared to when sated under both pair-housed and socially isolated conditions (Tables 1, 2). There was no effect of sex on body weight or food consumption measures (Tables 1, 2).

3.2 |. Experiment 1b: Social versus food preference was altered by food deprivation, but not social isolation in adult rats

Adult rats had a significant reduction in social over food preference scores when tested under food-deprived compared to sated conditions; neither housing condition nor sex altered stimulus preference (Table 3, Fig 3A, B). Even though social over food preference was reduced following food deprivation, adult rats had a strong preference for the social stimulus under both sated (pair-housed: t₍₁₅₎=10.7,p <0.001, d = 2.68; socially isolated: t₍₁₅₎ = 10.1, p < 0.001, d = 2.52), and food-deprived (pair-housed: t₍₁₅₎ = 3.81, p = 0.002, d = 0.95; socially isolated: t₍₁₅₎ = 4.35, p = 0.001, d = l.09; Fig 3A, B) conditions.

Table 3.

ANOVA statistics and partial eta squared (η²) effect sizes for Experiment 1b: Social versus food preference was altered by food deprivation, but not social isolation in adult rats. Significant effects shown in bold, n.s. = none significant, n.a. = not applicable, BW = body weight.

	Sex	Hunger Condition	Housing Condition	Interactions
Social over Food Preference [%]	F(1,14) = 0.12, p = 0.73, η2 = 0.009	F(1,14) = 8.83, p = 0.010, η2 = 0.39	F(1,14) = 0.08, p = 0.78, η2 = 0.006	n.s.
Social Stimulus Investigation [sec]	F(1,14) = 2.27, p = 0.15, η2 = 0.14	F(1,14) = 2.48, p = 0.14, η2 = 0.15	F(1,14) = 0.00, p = 0.99, η2 = 0.99	Hunger × Housing: F(1,14) = 5.62, p = 0.033, η2 = 0.29
Food Stimulus Investigation [sec]	F(1,14) = 1.24, p = 0.28, η2 = 0.082	F(1,14) = 16.1, p = 0.001, η2 = 0.53	F(1,14) = 0.02, p = 0.89, η2 = 0.001	n.s.
Total Investigation [sec]	F(1,14) = 3.57, p = 0.080, η2 = 0.20	F(1,14) = 21.8, p < 0.001, η2 = 0.61	F(1,14) = 0.005, p = 0.95, η2 < 0.001	Hunger × Housing: F(1,14) = 5.21, p = 0.039, η2 = 0.27
Total Distance [m]	F(1,14) = 4.22, p = 0.070, η2 = 0.32	F(1,9) = 0.03, p = 0.87, η2 = 0.003	F(1,9) = 1.36, p = 0.27, η2 = 0.13	n.s.
Middle Chamber Entries [#]	F(1,14) = 0.06, p = 0.81, η2 = 0.004	F(1,14) = 0.47, p = 0.50, η2 = 0.033	F(1,14) = 2.06, p = 0.17, η2 = 0.13	Sex × Hunger × Housing: F(1,14) = 5.34, p = 0.037, η2 = 0.28
Body Weight [% change]	F(1,14) = 8.89, p = 0.010, η2 = 0.39	F(1,14) = 298.7, p < 0.001, η2 = 0.96	F(1,14) = 0.24, p = 0.63, η2 = 0.017	n.s.
Consumption [% of BW] - Pair-Housed	F(1,6) = 0.94, p = 0.37, η2 = 0.14	F(1,6) = 67.5, p < 0.001, η2 = 0.92	n.a.	n.s.
Consumption [% of BW] - Socially Isolated	F(1,14) = 1.14, p = 0.30, η2 = 0.075	F(1,14) = 30.9, p < 0.001, η2 = 0.69	n.a.	n.s.

Fig 3. Experiment 1b.

Food deprivation (A), but not social isolation (B), significantly attenuated preference for the social stimulus in adult rats. Food deprivation did not alter investigation of the social stimulus, but robustly increased investigation of the food stimulus (C). Representative heat maps of activity from one subject (D). Bar graphs (mean ± SEM) collapsed across sex for A and B (same data replotted to illustrate main effects), and across sex and housing condition for C; individual data collapsed across housing condition in C; * p < 0.05, **p < 0.01, *** p < 0.001 mixed-model ANOVA; ^ p < 0.05, one-sample t-Test from 50% (gray dashed line); n.s. = not significant; n = 8 males, n = 8 females.

When food-deprived, adult rats increased their investigation of the food stimulus resulting in an increase in total investigation time compared to when they were sated (Fig 3C, Table 3). There were no main effects of housing condition nor sex on stimulus investigation times in adult rats, however there were significant hunger condition by housing condition interactions on social and total investigation times (Fig 3C, Table 3). Post hoc comparisons showed that for both of these measures adult rats spent more time investigating under food-deprived compared to sated conditions when they had been socially isolated (social: sated = 58.1 ± 4.60 sec, food-deprived = 79.5 ± 7.27 sec, p = 0.028, total: sated = 73.3 ± 5.14 sec, food-deprived = 122 ± 9.53 sec, p = 0.001), but not when they had been pair-housed (social: sated = 68.3 ± 8.49 sec, food-deprived = 69.1 ± 7.70 sec, p = 0.91, total: sated = 89.4 ± 8.70 sec, food-deprived = 104 ± 10.1 sec, p = 0.09); no other paired comparisons reached significance (p > 0.05, all).

Locomotor activity, as measured by distance traveled, was similar across all conditions and between both sexes in adult rats (Tables 2, 3). However, there was a significant sex by hunger condition by housing condition interaction on locomotor activity as measured by middle chamber entries (Tables 2, 3). Post hoc comparisons showed that under food-deprived conditions adult males made fewer middle chamber entries when they had been pair-housed compared to when they had been socially isolated (p = 0.04), and that under pair-housed conditions adult males made more middle chamber entries when they were sated compared to when they were food-deprived (p = 0.025); no other paired comparisons reached significance (p > 0.05, all).

There was a significant main effect of hunger condition on percent change in body weight with adult rats losing weight under food-deprived but not under sated conditions (Tables 2, 3). There were no effects of housing condition on percent change in body weight, but the percent change in body weight was greater in adult females than adult males (Tables 2, 3). Under both pair-housed and socially isolated conditions, adult rats consumed significantly more food when food-deprived compared to when sated (Tables 2, 3). There was no effect of sex on food consumption (Tables 2, 3).

3.3 |. Experiment 2a: Social versus food preference was altered by food deprivation, but not social isolation in adolescent mice

Adolescent mice had a significant reduction in their social over food preference scores when tested under food-deprived compared to sated conditions; neither housing condition nor sex altered stimulus preference (Table 4, Fig 4A, B). Adolescent mice did not have a significant preference for either stimulus under sated conditions (pair-housed: t_(l4)= 0.45, p = 0.66, d = 0.12; socially isolated: t₍₁₄₎ = 1.78, p = 0.097, d = 0.46), and food deprivation resulted in the emergence of a food preference (pair-housed: t₍₁₄₎ = 15.18, p < 0.001, d = 3.92; socially isolated: t_(l4) = 12.14, p < 0.001, d = 3.13; Fig 4A, B).

Table 4.

ANOVA statistics and partial eta squared (η²) effect sizes for Experiment 2a: Social versus food preference was altered by food deprivation, but not social isolation in adolescent mice. Significant effects shown in bold, n.s. = none significant, BW = body weight.

	Sex	Hunger Condition	Housing Condition	Interactions
Social over Food Preference [%]	F(1,13) = 0.37, p = 0.55, η2 = 0.028	F(1,13) = 140, p < 0.001, η2 = 0.92	F(1,13) = 0.52, p = 0.48, η2 = 0.039	n.s.
Social Stimulus Investigation [sec]	F(1,13) = 4.10, p = 0.064, η2 = 0.24	F(1,13) = 7.90, p = 0.015, η2 = 0.38	F(1,13) = 0.76, p = 0.40, η2 = 0.055	Sex × Hunger × Housing: F(1,13) = 5.45, p = 0.036, η2 = 0.30
Food Stimulus Investigation [sec]	F(1,13) = 1.20, p = 0.29, η2 = 0.084	F(1,13) = 94.59, p < 0.001, η2 = 0.88	F(1,13) = 3.34, p = 0.091, η2 = 0.20	n.s.
Total Investigation [sec]	F(1,13) = 3.18, p = 0.10, η2 = 0.20	F(1,13) = 78.1,p <0.001, η2 = 0.86	F(1,13) = 4.46, p = 0.055, η2 = 0.26	n.s.
Total Distance [m]	F(1,9) = 1.40, p = 0.27, η2 = 0.14	F(1,9) = 0.81, p = 0.39, η2 = 0.082	F(1,9) = 5.55, p = 0.043, η2 = 0.38	n.s.
Middle Chamber Entries [#]	F(1,13) = 0.006, p = 0.94, η2 < 0.001	F(1,13) = 4.54, p = 0.053, η2 = 0.26	F(1,13) = 1.34, p = 0.27, η2 = 0.093	n.s.
Body Weight [% change]	F(1,13) = 0.015, p = 0.90, η2 = 0.001	F(1,13) = 1,13, p < 0.001, η2 = 0.99	F(1,13) = 8.69, p = 0.011, η2 = 0.40	n.s.
Consumption [% of BW]	F(1,13) = 15.1, p = 0.002, η2 = 0.54	F(1,13) = 344.0, p < 0.001, η2 = 0.96	F(1,13) = 2.64, p = 0.13, η2 = 117	Sex × Hunger F(1,13) = 8.63, p = 0.012, η2 = 0.40

Fig 4. Experiment 2a.

Food deprivation (A), but not social isolation (B), significantly altered stimulus preference in adolescent mice which resulted in the emergence of a food preference under food-deprived conditions. Food deprivation mildly decreased investigation of the social stimulus and robustly increased investigation of the food stimulus (C). Representative heat maps of activity from one subject (D). Bar graphs (mean ± SEM) collapsed across sex for A and B (same data replotted to illustrate main effects), and across sex and housing condition for C; individual data collapsed across housing condition in C; * p < 0.05, ** p < 0.01, *** p < 0.001 mixed-model ANOVA; ^ p < 0.05, one-sample t-Test from 50% (gray dashed line); n.s. = not significant; n = 8 males, n = 7 females.

The observed changes in stimulus preference were due to a mild, but significant, decrease in the time spent investigating the social stimulus and a robust increase in the time spent investigating the food stimulus which increased total investigation time when adolescent mice were food-deprived compared to when they were sated (Table 4, Fig 4). Food and total investigation times were unaffected by housing condition or sex in adolescent mice, but there was a significant sex by hunger condition by housing condition interaction on social investigation times (Table 4). Post hoc comparisons showed that pair-housed adolescent female mice investigated the social stimulus less when they were food-deprived compared to when they were sated (food-deprived: 26.1 ± 4.69, sated: 42.5 ± 4.13, p = 0.011), and that when subjects were tested under pair-housed and sated conditions adolescent male mice investigated the social stimulus less than adolescent female mice (males: 25.1 ± 3.87 sec, females: 42.5 ± 4.13 sec, p = 0.009); no other paired comparisons reached significance (p > 0.05, all).

Adolescent mice had a tendency to make more middle chamber entries when sated compared to when food-deprived, but distance traveled was similar across hunger conditions (Tables 2, 4). Adolescent mice traveled further under pair-housed compared to socially isolated conditions, but middle chamber entries were similar across housing conditions (Tables 2, 4). There was no effect of sex on either locomotor activity measure in adolescent mice (Tables 2, 4).

As expected, adolescent mice lost weight under food-deprived conditions and gained weight under sated conditions resulting in a significant main effect of hunger condition on percent change in body weight (Tables 2, 4). Interestingly, there was also a significant main effect of housing condition on percent change in body weight; adolescent mice lost more weight when food-deprived and gained less weight when sated when tested under socially isolated conditions compared to under pair-housed conditions (Tables 2, 4). There was no effect of sex on percent change in body weight in adolescent mice (Tables 2, 4). As expected, adolescent mice consumed significantly more food when food-deprived compared to when sated (Tables 2, 4). There was a significant sex by hunger condition interaction on food consumption (Tables 2, 4). Post hoc comparisons showed that adolescent female mice consumed more food relative to their body weight than adolescent male mice when subjects were food-deprived (p = 0.002); the sexes consumed a similar amount of food when subjects were sated (p = 0.43). There was no effect of housing condition on food consumption in adolescent mice (Tables 2, 4).

3.4 |. Experiment 2b: Social versus food preference was altered by food deprivation, but not social isolation in adult mice

Adult mice had a significant reduction in their social over food preference scores when tested under food-deprived compared to sated conditions; neither housing condition nor sex altered stimulus preference (Table 5, Fig 5A, B). Adult mice did not have a significant preference for either stimulus under sated conditions (pair-housed: t₍₁₄₎ = 0.61, p = 0.55, d = 0.16; socially isolated: t₍₁₄₎ = 1.44, p = 0.17, d = 0.37) while food deprivation resulted in the emergence of a food preference (pair-housed: t₍₁₄₎ = 9.09, p < 0.001, d = 2.35; socially isolated: t₍₁₄₎ = 5.26, p < 0.001, d = 1.36; Fig 5A, B).

Table 5.

ANOVA statistics and partial eta squared (η²) effect sizes for Experiment 2b: Social versus food preference was altered by food deprivation, but not social isolation in adult mice. Significant effects shown in bold, n.s. = none significant, BW = body weight.

	Sex	Hunger Condition	Housing Condition	Interactions
Social over Food Preference [%]	F(1,13) = 1.94, p = 0.19, η2 = 0.13	F(1,13) = 44.8, p < 0.001, η2 = 0.78	F(1,13) = 0.014, p = 0.91, η2 = 0.001	n.s.
Social Stimulus Investigation [sec]	F(1,13) = 1.86, p = 0.20, η2 = 0.13	F(1,13) = 0.67, p = 0.43, η2 = 0.049	F(1,13) = 0.37, p = 0.56, η2 = 0.027	n.s.
Food Stimulus Investigation [sec]	F(1,13) = 0.008, p = 0.93, η2 = 0.001	F(1,13) = 41.1, p < 0.001, η2 = 0.76	F(1,13) = 0.47, p = 0.51, η2 = 0.035	n.s.
Total Investigation [sec]	F(1,13) = 0.18, p = 0.68, η2 = 0.014	F(1,13) = 43.5, p < 0.001, η2 = 0.77	F(1,13) = 1.43, p = 0.25, η2 = 0.099	n.s.
Total Distance [m]	F(1,12) = 5.98, p = 0.03, η2 = 0.33	F(1,12) = 7.90, p = 0.016, η2 = 0.40	F(1,12) = 1.32, p = 0.27, η2 = 0.099	n.s.
Middle Chamber Entries [#]	F(1,13) = 12.89, p = 0.003, η2 = 0.50	F(1,13) = 17.4, p = 0.001, η2 = 0.57	F(1,13) = 0.06, p = 0.82, η2 = 0.004	Hunger × Sex: F(1,13) = 4.72, p = 0.049, η2 = 0.27
Body Weight [% change]	F(1,13) = 0.004, p = 0.95, η2 < 0.001	F(1,13) = 272, p < 0.001, η2 = 0.95	F(1,13) = 0.36, p = 0.56, η2 = 0.027	Hunger × Housing: F(1,13) = 18.4 p = 0.001, η2 = 0.59
Consumption [% of BW]	F(1,13) = 1.01, p = 0.33, η2 = 0.072	F(1,13) = 147, p < 0.001, η2 = 0.92	F(1,13) = 0.31, p = 0.59, η2 = 0.023	n.s.

Fig 5. Experiment 2b.

Food deprivation (A), but not social isolation (B), significantly altered stimulus preference in adult mice, which resulted in the emergence of a food preference under food-deprived conditions. Food deprivation robustly increased investigation of the food stimulus, but did not change investigation of the social stimulus (C). Representative heat maps of activity from one subject (D). Bar graphs (mean ± SEM) collapsed across sex for A and B (same data replotted to illustrate main effects), and across sex and housing condition for C; individual data collapsed across housing condition in C; *** p < 0.001 mixed-model ANOVA; ^ p < 0.05, one-sample t-Test from 50% (gray dashed line); n.s. = not significant; n = 8 males, n = 7 females.

The observed changes in stimulus preference were due to a robust increase in the time spent investigating the food stimulus, which increased total investigation time when adult mice were food-deprived compared to when they were sated; time spent investigating the social stimulus was similar across conditions (Table 5, Fig 5). Investigation times were unaffected by housing condition or sex in adult mice (Table 5).

There was a main effect of sex on locomotor activity as measured by distance traveled; adult female mice traveled further than adult male mice (Tables 2, 5). There was also a main effect of hunger condition on distance traveled; adult mice traveled further when sated compared to when they were food-deprived (Tables 2, 5). There was a significant sex by hunger condition interaction on locomotor activity as measured by middle chamber entries, but not by distance traveled. Post hoc paired comparisons showed that adult female mice (p = 0.001), but not adult male mice (p = 0.17), made more chamber entries when sated compared to when food-deprived. Further, similar to distance traveled, adult female mice made more middle entries than adult male mice when sated (p = 0.001) and when food-deprived (p = 0.024). Locomotor activity measures were unaffected by housing condition in adult mice (Tables 2, 5).

As expected, there was a significant main effect of hunger condition on percent change in body weight; adult mice lost weight under food-deprived compared to under sated conditions (Tables 2, 5). There was also a significant hunger condition by housing condition interaction on percent change in body weight (Tables 2, 5). Post hoc comparisons showed that when adult mice were socially isolated compared to when they were pair-housed, they lost more weight when food-deprived (p = 0.018), and gained more when they were sated (p = 0.001). These patterns were similar between males and females; there was no effect or interaction with sex on percent change in body weight (Tables 2, 5). Adult mice consumed significantly more food relative to their body weight when food-deprived compared to when sated; there were no effects of sex or housing condition on food consumption (Tables 2, 5).

3.5 |. Experiment 3a: Social over food preference was greater for a novel social stimulus compared to a familiar social stimulus in adolescent rats

Adolescent rats had higher social over food preference scores when the social stimulus was novel compared to when the social stimulus was their cagemate (Fig 6A, Table 6); there was no effect of sex on stimulus preference. Adolescent rats exhibited a robust social preference, and this was true for both the novel (t_(l6) = 14.28, p < 0.001, d = 3.46) and familiar (cagemate; t_(l6) = 7.04, p < 0.001, d = 1.71) social stimuli (Fig 6A).

Fig 6. Experiments 3a and 3b.

Adolescent rats exhibited a robust preference for the social stimulus, which was attenuated when tested with their cagemate compared to when tested with a novel social stimulus (A). Adolescent rats spent more time investigating the social stimulus when it was novel than when it was their cagemate; food investigation was low and similar across conditions (B). Adolescent mice exhibited a preference for the food stimulus when the social stimulus was their cagemate (C, right), but there were no significant changes in preference (C, left) or investigation times (D) between social stimulus conditions. Bar graphs (mean ± SEM) collapsed across sex; ** p < 0.01, *** p < 0.001, mixed-model ANOVA; ^ p < 0.05, one-sample t-Test from 50% (gray dashed line); n.s. = not significant; rats: n = 9 males, n = 8 females; mice: n = 8 males, n = 8 females.

Table 6.

ANOVA statistics and partial eta squared (η²) effect sizes for Experiment 3a: Social over food preference was greater for a novel social stimulus compared to a familiar social stimulus in adolescent rats. Significant effects shown in bold, n.s. = none significant.

	Sex	Social Saliency	Interaction
Social over Food Preference [%]	F(1,15) = 0.76, p = 0.40, η2 = 0.048	F(1,15) = 20.9, p < 0.001, η2 = 0.58	n.s.
Social Stimulus Investigation [sec]	F(1,15) = 0.22, p = 0.65, η2 = 0.014	F(1,15) = 11.92, p = 0.004, η2 = 0.44	n.s
Food Stimulus Investigation [sec]	F(1,15) = 0.67, p = 0.43, η2 = 0.043	F(1,15) = 0.043, p = 0.84, η2 = 0.003	n.s.
Total Investigation [sec]	F(1,15) = 0.10, p = 0.75, η2 = 0.007	F(1,15) = 8.99, p = 0.009, η2 = 0.38	n.s.
Total Distance [m]	F(1,15) = 1.06, p = 0.32, η2 = 0.066	F(1,15) = 0.03, p = 0.87, η2 = 0.002	n.s.
Middle Chamber Entries [#]	F(1,15) = 0.08, p = 0.79, η2 = 0.005	F(1,15) = 2.33, p = 0.15, η2 = 0.13	n.s.

Adolescent rats spent significantly more time investigating the social stimulus when it was novel compared to when it was their cagemate which resulted in a significantly greater total stimulus investigation time when the social stimulus was novel; investigation of the food stimulus was low and similar across conditions (Table 6, Fig 6B). There were no sex differences in social, food, or total stimulus investigation times in adolescent rats (Table 6, Fig 6B).

Neither sex nor social saliency condition affected locomotor activity as measured by total distance traveled or middle chamber entries in adolescent rats (Tables 6, 7).

Table 7.

Activity measures for Experiments 3a and 3b. Data shown as mean ± SEM; ¹main effect of sex; see text for details and Tables 6 and 8 for corresponding ANOVA statistics.

		Cagemate	Novel
Total Distance Traveled [m]
Exp 4a: Adolescent Rats	Males	27.0 ± 2.05	24.6 ± 1.64
	Females	28.3 ± 4.70	30.0 ± 2.05
Exp 4b: Adolescent Mice1	Males	18.6 ± 1.36	18.2 ± 1.94
	Females	23.0 ± 0.83	24.8 ± 1.73
Middle Chamber Entries [#]
Exp 4a: Adolescent Rats	Males	28.7 ± 2.11	23.1 ± 1.59
	Females	25.4 ± 4.64	24.5 ± 2.58
Exp 4b: Adolescent Mice1	Males	18.9 ± 2.09	21.1 ± 2.56
	Females	25.9 ± 1.36	30.9 ± 4.59

3.6 |. Experiment 3b: Stimulus preference was altered by social familiarity in adolescent mice

Adolescent mice did not show a significant within-subjects change in stimulus preference as a result of social saliency (Fig 6C, Table 8), and stimulus preference was similar between males and females. However, while adolescent mice did not have a stimulus preference when the social stimulus was novel (t₍₁₅₎ = 0.099, p = 0.92, d = 0.025; Fig 6C, right), they exhibited a significant preference for the food stimulus when the social stimulus was their cagemate (t₍₁₅₎ = 2.48, p = 0.026, d = 0.62; Fig 6C, left).

Table 8.

ANOVA statistics and partial eta squared (η²) effect sizes for Experiment 3b: Stimulus preference was altered by social familiarity in adolescent mice. Significant effects shown in bold, n.s. = none significant.

	Sex	Social Saliency	Interaction
Social over Food Preference [%]	F(1,14) = 0.006, p = 0.94, η2 < 0.001	F(1,14) = 1.86, p = 0.20, η2 = 0.12	n.s.
Social Stimulus Investigation [sec]	F(1,14) = 0.44, p = 0.52, η2 = 0.030	F(1,14) = 2.39, p = 0.14, η2 = 0.15	n.s.
Food Stimulus Investigation [sec]	F(1,14) = 0.58, p = 0.46, η2 = 0.040	F(1,14) = 0.012, p = 0.92, η2 = 0.001	n.s.
Total Investigation [sec]	F(1,14) = 1.91, p = 0.19, η2 = 0.12	F(1,14) = 0.63, p = 0.44, η2 = 0.043	n.s.
Total Distance [m]	F(1,13) = 8.03, p = 0.014, η2 = 0.38	F(1,14) = 0.57, p = 0.46, η2 = 0.042	n.s.
Middle Chamber Entries [#]	F(1,14) = 6.37, p = 0.024, η2 = 0.31	F(1,14) = 2.22, p = 0.16, η2 = 0.14	n.s.

There was no effect of sex nor social saliency condition on social, food, or total investigation times in adolescent mice (Fig 6D, Table 8).

Locomotor activity was higher in adolescent female mice compared to adolescent male mice as measured by total distance traveled and middle chamber entries, but was unaffected by social saliency condition (Tables 7, 8).

3.7 |. Experiment 4a: Adolescent mice preferred to investigate a novel social stimulus over an empty corral

Adolescent mice spent significantly more time investigating the corralled social stimulus compared to the empty corral (F_(1,6) = 7.10, p = 0.037, η² = 0.54; Fig 7A). There were no effects of sex on investigation times (F_(1,6) = 0.38, p = 0.56, η² = 0.060; Fig 7A) or stimulus preference (t₍₆₎ = 0.65, p = 0.54, d = 0.46; Fig 7B). Adolescent mice exhibited a significant preference for the social stimulus (t₍₇₎ = 2.91, p = 0.023, d = 1.03; Fig 7B).

Fig 7. Experiments 4a and 4b.

Adolescent (A) and adult (C) mice spent more time investigating a corral containing a novel social stimulus than an empty corral, and as a result showed a significant preference for the corral containing the social stimulus (B, adolescents; D, adults). Bar graphs (mean ± SEM) collapsed across sex; * p < 0.05 mixed-model ANOVA; ^ p < 0.05, one-sample t-Test from 50% (gray dashed line); adolescents: n = 4 males, n = 4 females; adults: n = 4 males, n = 4 females).

There were no sex differences in middle chamber entries (males: 32 ± 2.3, females: 37 ± 3.1; t₍₆₎ = 1.31, p = 0.24, d = 0.92), or total distance traveled (males: 29.1 ± 1.3 m, females: 25.4 ± 2.2 m; t₍₆₎ = 1.56, p = 0.17, d = 1.10) in adolescent mice.

3.8 |. Experiment 4b: Adult mice preferred to investigate a novel social stimulus over an empty corral

Adult mice spent significantly more time investigating the corralled social stimulus compared to the empty corral (F_(1,6) = 7.00, p = 0.038, η² = 0.54; Fig 7C). There were no effects of sex on investigation times (F_(1,6) = 1.95, p = 0.21, η² = 0.25; Fig 7C) or stimulus preference (t₍₆₎ = 0.21, p = 0.84, d = 0.15; Fig 7D). Adult mice exhibited a significant preference for the social stimulus (t₍₇₎ = 3.18, p = 0.016, d = 1.12; Fig 7D).

There were no sex differences in middle chamber entries (males: 17.8 ± 2.7, females: 24.5 ± 5.3; t₍₆₎ = 1.13,p = 0.30, d = 0.180), or total distance traveled (males: 15.8 ± 4.7 m, females: 20.2 ± 1.58 m; t₍₆₎ = 1.53, p = 0.18, d = 0.46) in adult mice.

4 |. DISCUSSION

Here we characterized a behavioral paradigm designed to test the competition between the choice to seek social interaction versus the choice to seek food, and assessed how this competition was modulated by internal cues, external cues, sex, age, and rodent model. We demonstrated that changes in stimulus preference in response to our cue manipulations (i.e., food deprivation, social isolation, social salience; Fig 8C, F) were similar across cohorts (i.e., sex, age, and/or rodent model), but that baseline stimulus preference and investigation times varied greatly between Wistar rats and C57BL/6 mice. Specifically, Wistar rats were generally more social-preferring and C57BL/6 mice were generally more food-preferring (Fig 8A–B, D–E). Further, especially in Wistar rats, the degree of food deprivation-induced changes in investigation patterns appeared greater in adolescents compared to adults (Fig 8A–C). Given these cohort differences, our results highlight the importance of taking experimental population (i.e., age, rodent model) into account when using the Social versus Food Preference test in future experiments.

Fig 8.

Age and species differences in stimulus preference and investigation times in the Social versus Food Preference Test. Data has been replotted to visually compare the results between Experiments 1a, 1b, 2a, and 2b (A-C; the effects of social isolation and food deprivation; data collapsed across sex and housing condition), between Experiments 3a and 3b (D-F; the effect of social salience; data collapsed across sex), and between Experiments 4a and 4b (G-H; social stimulus versus an empty corral; data collapsed across sex). Across experiments and under baseline conditions, Wistar rats spent more time with the social stimulus than the food stimulus, while C57BL/6 mice spent equal time with the social and food stimuli (I). Additionally, across experiments and under baseline conditions, Wistar rats spent more time with social stimulus and less time with the food stimulus than C57BL/6 mice (I); A, B, D, E, G, H: data shown as mean ± SEM; C, F: black bars = mean.

4.1 |. The effects of internal cue manipulations on social versus food preference

To begin characterizing the Social versus Food Preference Test, we first determined how altering the internal motivational states of subjects by exposing them to acute social isolation and/or acute food deprivation affected stimulus preference and stimulus investigation times (Experiments 1, 2). In agreement with our predictions, adolescent and adult rats and mice of both sexes showed a significant reduction in their social over food preference scores in response to acute food deprivation (Fig 8C). This change was primarily driven by a significant increase in investigation of the food stimulus in response to food deprivation in all four cohorts (Fig 8B). This increased interest in the food stimulus corresponded with our proxy physiological and behavioral measures of hunger; acute food deprivation significantly decreased body weight and increased post-test food consumption compared to sated conditions in all four cohorts (Table 9).

Table 9.

Comparing the results across experiments for locomotor activity, body weight, and food consumption measures; n/a = not applicable; n.a. = not analyzed; n.d. = no differences between groups or conditions; M = males; F = females; S = sated; FD = food-deprived; PH = pair-housed; SI = socially isolated.

	Adolescent Rats	Adult Rats	Adolescent Mice	Adult Mice
Experiments 1a, 1b, 2a, & 2b: The effects of social isolation and food deprivation
Total Distance [m]	n.a.	n.d.	PH > SI	S > FD F > M
Middle Chamber Entries [#]	S > FD	PH-M: S > FD FD-M: SI > PH	S > FD	F: S > FD F > M
Body Weight [% change]	FD > S SI > PH	FD > S F > M	FD > S SI > PH	FD > S SI > PH
Consumption [% of BW]	FD > S	FD > S	FD > S FD: F > M	FD > S
Experiments 3a & 3b: The effect of social salience in sated and pair-housed adolescent rats and mice
Total Distance [m]	n.d.	n/a	F > M	n/a
Middle Chamber Entries [#]	n.d	n/a	F > M	n/a
Experiments 4a & 4b: Social stimulus versus an empty corral in sated and pair-housed mice
Total Distance [m]	n/a	n/a	n.d.	n.d.
Middle Chamber Entries [#]	n/a	n/a	n.d.	n.d.

4.1.1 |. Age differences in food deprivation-induced changes in stimulus preference and investigation times

While food deprivation decreased social over food preference scores in both adolescents and adults, the degree of change was different between the age groups (i.e., slope of lines in Fig 8A). Specifically, the mean difference in preference scores between sated and food-deprived conditions was greater in adolescents compared to adults for both rats and mice (Fig 8C). This age difference was partly due to differences in food deprivation-induced changes in social stimulus investigation in adolescents compared to adults. Adolescent rats and mice significantly reduced their investigation of the social stimulus when food-deprived compared to when sated, and this effect size was larger in rats (Tables 1, 4; Fig 8B). In contrast, adult rats and mice maintained a similar level of social stimulus investigation under both sated and food-deprived conditions (Tables 3, 5; Fig 8B). Further, while all cohorts significantly increased their investigation of the food stimulus under food-deprived compared to sated conditions (Fig 8B), the effect size was larger in adolescents compared to adults in both rats and mice (Tables 1, 3–5). These results suggest that adolescents may have been more affected by the food deprivation manipulation when compared to adults. In support, under food-deprived conditions adolescents lost, on average, a greater percent of their body weight, and consumed, on average, more food relative to their body weight compared to adults (Table 2).

In rats, we also observed age differences in the absolute amount of time subjects spent investigating the stimuli, and this varied by hunger state. Under sated conditions, adolescent rats spent, on average, 183% more time than adults investigating the social stimulus, and under food-deprived conditions adolescents spent, on average, 171 %more time than adult rats investigating the food stimulus (Fig 8B). Interestingly, investigation of the food stimulus was similar between adolescent rats and adult rats under sated conditions suggesting similar baseline levels in food motivation between the ages, while food deprivation reduced the time adolescent rats spent investigating the social stimulus down to adult rat levels (Fig 8B). These age differences are in alignment with increased reward-seeking behavior and motivation for a variety of rewards in adolescents compared to adults, characteristics believed to be evolutionarily important for typical development [for review see: 17, 18]. Since this adolescent-specific peak in reward-seeking and motivated behaviors has been observed across species, including mice, it was surprising that we observed similar amounts of stimulus investigation in adolescent versus adult mice (Fig 8B). One reason could be the age of testing for the adolescent mice. Indeed, an earlier study reporting greater levels of social investigation by adolescent compared to adult mice [19], tested adolescent mice at a younger age (30–32 days-old) than we did in the current study (37–44 days-old). Thus, although we attempted to equilibrate the adolescent time-point between species in our experiments [15], differences may exist between rats and mice in the developmental time course of increased reward-seeking behaviors.

4.1.2 |. Species differences in baseline stimulus preference and investigation times

While social over food preference scores in response to food deprivation were similarly decreased in Wistar rats and C57BL/6 mice, baseline preference scores were vastly different between these two rodent models. Wistar rats exhibited a strong social preference when tested under sated conditions and this was abolished (in adolescents) or attenuated (in adults) by food deprivation (Fig 8A). In contrast, adolescent and adult C57BL/6 mice had no stimulus preference when tested under sated conditions and had a strong food preference when food-deprived (Fig 8A). The adult mouse results partially matched a study where adult male mice preferred a food stimulus over a social stimulus when food-deprived, but preferred the social stimulus over the food stimulus when sated [3]. Methodological differences between the two studies may account for the differing results under sated conditions, specifically, the length of social isolation (18 hrs in our study vs. > 2.5 weeks in [3]), type of social stimulus (age-, and sex-matched in our study vs. receptive female or juvenile male in [3]), food presentation (corralled in our study vs. available for consumption in [3]), and/or chamber design (3 chambers which provided a neutral zone in our study vs. 2 chambers which forced choice in [3]). Our adult mouse results are also consistent with a study utilizing a homecage resident-intruder paradigm where single-housed adult male mice increased food intake (standard laboratory chow) and decreased mating (receptive female intruder) but not aggressive (subordinate male intruder) behaviors under food-deprived (18 or 48 hr) compared to sated conditions [8].

The distinct difference in stimulus preference between Wistar rats and C57BL/6 mice, which was also observed in our experiments investigating the role of social salience (Fig 8D, Section 4.2.2 below), was driven by differences in the time spent investigating both the social and the food stimuli. Collapsed across experiment (Experiments 1–3), age (adolescents, adults), sex (males, females), and manipulation (hunger, housing, and social salience conditions), Wistar rats spent, on average, 243 %more time investigating the social stimulus than C57BL/6 mice, and C57BL/6 mice spent, on average, 240 %more time investigating the food stimulus than Wistar rats (Fig 8B, E). Thus, our results support the hypothesis that rats and mice differ in their baseline motivation to seek social interaction (higher in rats) and to seek food (higher in mice). This may be due to differences in the evolutionary or ecological history of these rodent species, and corresponds with multiple lines of evidence suggesting that, socially, rats are more agonistic and mice are more antagonistic [for reviews see: 20, 21]. Further, our results are in agreement with a prior study that compared 6–8 week-old male Sprague-Dawley rats and C57BL/6N mice, and found that rats were more likely than mice to develop a social conditioned place preference and that rats spent longer interacting with the social stimulus than mice during the initial conditioning sessions [22]. Although unlikely, it should be noted that differences in the size, construction materials, and/or corral shape between our rat-sized and mouse-sized equipment could have contributed to the observed species differences in stimulus preference and investigation times.

Differences in the choice to seek social interaction versus food could also be influenced by species differences in metabolism [23, 24]. Indeed, although the length of food deprivation used in the current study was longer for rats than mice (24 hrs versus 18 hrs), mice lost, on average, a greater percent of their body weight in response to food deprivation than rats suggesting mice have higher metabolic demands (Table 2). Lastly, it is important to recognize that only one strain of each species was compared in the present study, and that within-species strain differences exist in other social [e.g., 11, 25, 26–29] and food-related behavioral assays [e.g., 30, 31–34]. Further, because our aim was to compare two of the most commonly used rodent models [35], we used an outbred rat strain and an inbred mouse strain which could have also influenced our findings. Thus, future experiments utilizing additional inbred and outbred strains of each species are needed to determine if the results from the present study are reflective of each species as a whole.

4.1.3 |. Males and females exhibited similar stimulus preferences and investigation times

No sex differences in stimulus preference or stimulus investigation times were observed in any of the experiments in the current study. Within each cohort, males and females exhibited similar patterns of stimulus preference and similar levels of stimulus investigation in the Social versus Food Preference Test under all examined conditions. Nevertheless, we believe it is important to use both sexes in future investigations of the neural substrates underlying social versus food preference [36, 37]. This is critical because an absence of sex differences in behavior could still be caused by sex differences in its neural underpinnings [38], and we and others have demonstrated that the neural mechanisms underlying motivated social behavior can differ in males and females [39–45].

4.1.4 |. Social isolation did not affect stimulus investigation patterns, but did potentiate food deprivation-induced weight loss

Contrary to our prediction, acute social isolation did not affect stimulus preference or stimulus investigation in any cohort (Fig 8A–C). Our prediction was based on prior work reporting increased social investigation by rats and mice following short-term (3–24 hr) social isolation [e.g., 12, 46, 47, 48]. Additionally, a recent meta-analysis found that independent of age or sex, rats and mice consume more food when socially isolated compared to when socially housed [49] suggesting that social isolation may also increase food-directed motivation. It should be noted, however, that the minimum amount of social isolation examined in this meta-analysis (48 hr) was at least twice that used in our current study (24 hrs in rats, 18 hrs in mice). A social isolation-induced increase in food-directed motivation may be an adaptive response since socially isolated subjects are unable to engage in huddling, a behavior commonly expressed by rats and mice [50–52] and which is thought to serve a thermoregulatory purpose [53]. Thus, these long-term socially isolated subjects may have had increased metabolic demands that manifested as increased food-directed motivation. The shorter length of social isolation used in our current study may not have been long enough to induce measurable changes in social- or food-directed motivation and subsequent behavior in the Social versus Food Preference Test, despite prior evidence that short-term social isolation increases social investigation [e.g., 12, 46, 47, 48]. In support, an earlier study found that sated adult male mice that had been isolated for at least 2.5 weeks preferred a social stimulus (receptive female or juvenile male) over a food stimulus (standard laboratory chow) [3]. Assessments of post-test social investigation in the homecage, analogous to the post-test food consumption tests conducted in the current studies, would be one way examine the effects of the social isolation manipulation in future experiments. Further, we equated the absolute duration of food deprivation and social isolation in our experiments (i.e., 24 hrs for both manipulations in rats, 18 hrs for both manipulations in mice), but this may not elicit equal changes in the motivation to seek social interaction compared to the motivation to seek food. Future studies could determine what length of social isolation would induce changes in behavior within this paradigm.

Social isolation did consistently alter one parameter examined in the current set of experiments: percent change in body weight (Table 9). Specifically, adolescent rats and adolescent and adult mice lost more weight under food deprivation when they were socially isolated compared to when they were pair-housed. Social isolation-induced changes in body weight were not observed in adult rats, which may reflect differences in metabolic demands corresponding to body size [24] or within-species age differences in spontaneous activity [54]. One likely explanation for increased body weight loss is the inability of socially isolated subjects to engage in huddling [50, 51, 53]. Thus, socially isolated subjects may have expended more energy for thermoregulation processes than pair-housed subjects, and in the absence of food were unable to compensate with an increase in energy intake. In future studies, conducting homecage behavioral observations and monitoring food consumption during the social isolation period could provide evidence to support this hypothesis.

4.2 |. The effects of external cue manipulation on social versus food preference

Since behavior is influenced both by internal and external cues [1, 2], we next sought to determine how behavior in the Social versus Food Preference Test would be affected by manipulation of external cues. Since adolescents showed greater changes in behavior compared to adults in our manipulations of internal cues (i.e., Experiments 1, 2; see Section 4.1.1), we chose to only use adolescents for these experiments. Specifically, we examined whether manipulating the saliency of the social stimulus (i.e., novel versus familiar) would alter social versus food preference in sated adolescent rats and mice of both sexes (Experiment 3).

Given the previously described preference of rats and mice for social novelty [10–12, 25, 26, 55], and the increased novelty-seeking behavior displayed by adolescents across species [for review see: 17, 18], we predicted that social novelty, as opposed to social familiarity, would bias preference more towards the social stimulus in the Social versus Food Preference Test. In general, our prediction was confirmed; adolescent rats displayed a stronger preference for the social stimulus when the social stimulus was a novel conspecific compared to when it was their cagemate, and while adolescent mice did not have a stimulus preference when the social stimulus was a novel conspecific they preferred the food stimulus when the social stimulus was their cagemate (Fig 8D). These results further reinforce our working hypothesis that while the baseline balance between social motivation and food motivation differs between rats and mice (see Section 4.1.2), how rats and mice respond to experimental manipulations of these motivations is similar.

4.3 |. Despite a lack of social over food preference, C57BL/6 mice prefer a social stimulus over an empty corral

Prior studies on general sociability have consistently shown that C57BL/6 mice prefer a social stimulus over an object [25, 26] or empty chamber [11]. In agreement with these prior studies, we found that adolescent and adult male and female C57BL/6 mice showed a significant preference for a corralled social stimulus versus an empty corral. When considered with the results from Experiments 2 and 3b this suggests that in C57BL/6 mice tested under sated conditions the motivation to investigate corralled a food stimulus is similar to the motivation to investigate a corralled social stimulus, and both are greater than the motivation to investigate an empty corral. However, it should be noted this refers to motivation as measured by choice in a passive investigation paradigm. In an operant two-choice lever-pressing paradigm, sated adult male C57BL/6J and BTBR T+tf/J mice had greater motivation for a palatable food reward (sweetened evaporated milk) compared to a social reward (sex- and age-matched stimulus mouse) [34]. Whether the difference in the balance between social motivation and food motivation between this prior study and the current study was due to differences in the nature of the task (operant in [34] versus passive investigation in current study) or the value of the food stimulus (palatable in [34] versus standard laboratory chow in current study) could be addressed in future studies by using a palatable food stimulus in the Social versus Food Preference Test.

4.4 |. The Social versus Food Preference Test is a flexible behavioral paradigm for studying competing motivations

The present series of experiments highlight the flexible nature of the Social versus Food Preference Test, and support its use in future studies of motivated behavioral choice and interrogations of the underlying peripheral and central systems. A key benefit of this behavioral test is the ability to assess both absolute (i.e., investigation time) and relative (i.e., preference) interest to investigate two opposing stimuli, giving multiple readouts for interpreting how manipulations of test parameters, or peripheral or central systems, can affect behavior. Further, the ease of manipulating internal motivational states (e.g., varying the length of food deprivation or social isolation), will allow experimenters to titer stimulus preference or investigation times. When combined with careful consideration of experimental population, this ability will be key in tailoring experiments for specific hypotheses and preventing floor or ceiling effects (e.g., prediction for decreased social preference: select sated Wistar rats; prediction for decreased food preference: select food-deprived C57BL/6 mice; uncertain direction of prediction: select sated C57BL/6 mice or food-deprived Wistar rats). Moreover, the numerous possibilities for manipulating the salience, reward value, or availability of the social (e.g., novelty, age, sex, tethered instead of corralled) or food stimuli (e.g., novelty, palatability, stimulus devaluation, previous pairing with LiCl, free-access instead of corralled), will also allow experimenters to titer stimulus preference or investigation times and address an even larger set of hypotheses regarding how subjects choose which stimulus type to investigate.

The version of the Social versus Food Preference Test used in the experiments presented here was adapted from a two-chamber social interaction assay assessing the effects of hunger signals on social interest in mice [3], and served to validate the use of such a paradigm to investigate the choice of subjects to interact with a social versus a food stimulus. One methodological difference between the current and prior paradigms was the availability of the food stimulus (corralled to prevent consumption in current study, secured to the apparatus with adhesive putty which allowed for food consumption in the prior study [3]). The expression of motivated behaviors is classically broken down into appetitive (i.e., initial, seeking) and consummatory (i.e, final, satisfaction of drive) behaviors [56], and corralling both stimuli narrows the analyses to appetitive behaviors in the present version of the Social versus Food Preference Test. This enhances the ability to directly compare interest for each of the two stimuli because the dependent measure (i.e., investigation time) is the same allowing for the calculation of a preference score, where it would be more difficult to directly compare how the duration or frequency of social behaviors (e.g., anogenital investigation, social play, aggression, mating) compares with food-directed behaviors (e.g., bout frequency, bout duration, consumption). In order to dissociate appetitive and consummatory motivation, other groups have utilized operant paradigms where subjects work to gain access to a social or a food stimulus [6, 34] which allows for direct comparison of appetitive responding for each stimulus as well as quantification of subsequent consummatory behaviors. Another, perhaps more ecologically valid, approach to studying social versus food motivation was demonstrated using a homecage resident-intruder paradigm [8]. This paradigm provides the additional benefit of observing food-directed behaviors by the social stimulus and how these might impact, or be impacted by, the behavior of the experimental subject. Determining which type of paradigm to use for future investigations of social versus food behavioral choice should depend on the specific research question(s) to be addressed.

5 |. CONCLUSIONS

Here we established the Social versus Food Preference Test to examine the competition between the choice to seek social interaction versus the choice to seek food. First, we characterized how this competition was modulated by internal cues (i.e., acute food deprivation and/or acute social isolation) and assessed whether these manipulations would produce similar changes in behavior between the sexes (males, females), across the lifespan (adolescents, adults), and between commonly used laboratory rodent models (Wistar rats, C57BL/6 mice). We found that behavior in this test was similar between the sexes and unaffected by social isolation, but highly influenced by food deprivation (i.e., biased preference more towards the food stimulus) and that this effect size was larger in adolescents than adults. Most strikingly, we observed a robust baseline difference in stimulus preference between Wistar rats and C57BL/6 mice: Wistar rats were generally more social-preferring and C57BL/6 mice were generally more food-preferring. Next, we determined whether the competition between the choice to seek social interaction versus the choice to seek food could be modulated by external cues and found that behavior was altered by changing the salience of the social stimulus (i.e., social novelty biased preference more towards the social stimulus) in adolescent Wistar rats and C57BL/6 mice of both sexes. Together, our experiments confirm that the Social versus Food Preference Test is a flexible behavioral paradigm suitable for future interrogations of the peripheral and central systems that can coordinate the expression of stimulus preference related to multiple motivated behaviors.

HIGHLIGHTS.

• Rats prefer social over food when sated, and this is attenuated by food deprivation.

• Mice have no preference when sated, and prefer food over social when food-deprived.

• Rats prefer a familiar social stimulus or a novel social stimulus over food.

• Mice prefer food over a familiar social stimulus.

• Adolescent rats investigate social and food stimuli longer than adult rats.