Effects of Maternal Separation on Punishment-driven Risky Decision Making in Adolescence and Adulthood

Abstract

Early life adversity (ELA) is associated with a multitude of neural and behavioral aberrations. To develop treatments to mitigate the effects of ELA, it is critical to determine which aspects of cognition are affected and when these disturbances manifest across the lifespan. Here, we tested the effects of maternal separation, an established rodent model of ELA, on punishment-driven risky decision-making longitudinally in both adolescence (25-55 days old) and adulthood (80-100 days old). Risk-taking was assessed with the Risky Decision-making Task, wherein rats choose between a small, safe reward and a large reward accompanied by an escalating risk of punishment (foot shock). We observed that rats exposed to maternal separation were more prone to risk-taking than controls during adolescence, and demonstrated reduced latency to make both risky and safe decisions. Interestingly, this augmented risk-taking was no longer evident in adulthood. Males and females displayed comparable levels of risk-taking during adolescence then diverged in adulthood, with adult males displaying a sharp increase in risk-taking. Finally, we observed that risk-taking changed across the lifespan in rats exposed to maternal separation, but not in control rats. Collectively, these data reveal that ELA engenders risk-taking in adolescence but not adulthood, and that sex differences in risky decision-making are not evident until adulthood. This has important implications for the development of both behavioral and biological treatments to improve decision-making during the vulnerable adolescent period.

Introduction

Early life adversity (ELA) is associated with a multitude of neural and cognitive aberrations. A common form of ELA is sustained parental neglect, which predicts health problems, psychopathology, and social/financial hardship later in life (Chen et al., 2021; Matthews et al., 2022; Bowirrat et al., 2023). However, because incidence of childhood neglect is increased in marginalized individuals and those with low socioeconomic status (Reiss et al., 2019; Smith and Pollak, 2020; Howell et al., 2021), it is difficult to disentangle effects of childhood neglect from co-occurring sources of trauma. Problems with causality can be addressed using rodent models such as the maternal separation protocol (MSEP), wherein rat pups are separated from dams during early development (Ader et al., 1960). While rodent models have revealed that MSEP causes age- and sex-mediated effects on brain function and behavior (Andersen, 2015; Ellis and Honeycutt, 2021; Rincón-Cortés, 2023), many of the specific aspects of cognition altered by MSEP remain unclear.

Risky decision-making, or risk-taking, is defined as willingness to pursue rewards associated with adverse outcomes (Orsini et al., 2015a). While a moderate amount of risk-taking can be beneficial, excessive risk-taking can engender a range of deleterious physical, financial, and social consequences. ELA is associated with several disorders characterized by risk-taking, such as substance use disorder and gambling disorder (Lotzin et al., 2018; Loo et al., 2019; Bristow et al., 2022; al’Absi et al., 2023). Furthermore, MSEP in rodents has been shown to affect general motivation (Waters and Gould, 2022), the mesolimbic dopamine system (Jahng et al., 2010; González-Pardo et al., 2020; Hamdan et al., 2022; Rincón-Cortés, 2023), and prefrontal cortical areas/higher cognition (Brenhouse et al., 2013; Boutros et al., 2017; Honeycutt et al., 2020), all factors involved with ability to assess risk and reward. Despite correlational evidence and overlapping neuronal substrates, the causal relationship between MSEP and punishment-driven risky decision-making has yet to be investigated.

Here, we measured the effects of MSEP on punishment-driven risky decision-making at different developmental stages in male and female rats. Rats were exposed to MSEP or control conditions, then tested in the risky decision-making task (RDT). MSEP began at postnatal day (PD) 11, comparable to date of birth in humans (Romijn et al., 1991; Clancy et al., 2007), then culminated at PD21, approximately the beginning of early childhood (1-3 years old; Malinovskaya et al., 2018). Thus, the MSEP period modeled chronic isolation during the first year of human life. The RDT measures preference between small rewards and larger rewards associated with an escalating risk of foot shock punishment (Simon et al., 2009a; Simon and Setlow, 2012). Critically, preference for risky options in RDT is associated with a variety of addiction-relevant phenotypes, including impulsivity (Gabriel et al., 2019, 2023), cocaine and nicotine sensitivity (Mitchell et al., 2014; Gabriel et al., 2019; Orsini et al., 2020), cue salience (Olshavsky et al., 2014), and dopaminergic alterations in both dorsal and ventral striatum (Simon et al., 2011; Mitchell et al., 2014; Freels et al., 2020), suggesting that this form of risk-taking contributes to addiction vulnerability. Moreover, males are reliably more risk-prone than females in RDT (Orsini et al., 2016, 2021; Gabriel et al., 2023), which underscores the utility of this task for studying sex differences in decision-making.

We first tested rats in RDT during late adolescence (pd 25-60) in male and female rats exposed to MSEP or standard conditions. Then, we re-trained and tested animals at pd 70-85 to determine if effects persisted into adulthood. Finally, to confirm that any effects (or lack-thereof) of MSEP in adulthood were not limited to RDT with a specific punishment intensity, we tested rats on RDT with an increased shock amplitude (.2mA to.3mA).

Materials and Methods

Subjects:

Male and female Long-Evans rat breeders were obtained from Charles River Laboratories. Four litters were bred in house, with one pup from each litter switched with a sex matched pup from another to mitigate litter bias. Litters were then subjected to either maternal separation (n=18; 10F/8M) or standard care (n=16; 8F/8M) from PD11-PD20. At PD21, rats were weaned and individually housed throughout the duration of the experiments. All rats were maintained on a reversed 12-hour light/dark cycle with testing conducted during dark hours to ensure maximum activity. During testing, rats were maintained at 90% of their free-feeding weight and ad libitum access to water. During adolescence, this was based on Long-Evans adolescent growth curves for males and females obtained from Charles River. Animal procedures were approved by the University of Memphis Institutional Animal Care and Use Committee.

Materials:

Transparent plastic cages with wire top grates, bedding material, and kimwipes for enrichment were used to house all rats. For the maternal separation procedure, comparable individual cages were used to house pups while separated from the dam. After completion of MSEP, animals were pair housed with a littermate. Standard operant chambers (Med Associates) were used to measure risky decision-making. Each chamber contained 2 retractable levers, an illuminable food trough, nose-poke apparatus, pellet dispenser, and shock grate. Standard dustless sucrose pellets (Bio-serv) were used as behavioral reinforcers.

Overall Experimental Procedure:

Rats were exposed to either maternal separation or standard conditions from PD11-21, then were weaned and pair-housed. This early period was chosen for two main reasons: to model isolation in the first year of life (Malinovskaya et al., 2018), and provide sufficient time for subjects to learn RDT (typically 30-40 days). Food restriction began on PD22. Rats began shaping procedures for RDT on PD25, then began RDT on ~PD30. Rats ran RDT for a minimum of 20 sessions, culminating on PD60. Rats were then given a 10-day rest with free access to food, then were again restricted to 90% weight and retrained on RDT in adulthood (PD70-85). Then, rats performed RDT with shock intensity increased from .2 to .3mA. from PD86-93. See Figure 1 for full visual timeline.

Figure 1:

Timeline of maternal separation and risky decision-making experiments

Maternal Separation Protocol:

Maternal separation was modified from Honeycutt et al (2020). Beginning at PD11, pups were separated into individual cages without food or water in a separate room for 4 hours for 10 consecutive days. Pups were weighed every other day throughout this protocol to confirm maintenance of a healthy weight. Control rats remained with their dam and litter 24 hours a day throughout this period. To minimize litter bias, one pup from each litter was swapped with another in the opposite condition. Any health issues, such as impaired movement, extreme weight loss, or aggression resulted in the animal in question being removed from the experiment.

Shaping Procedures

Prior to RDT, rats performed shaping procedures described in detail in Orsini and Simon (2020). Rats were first given a single session of magazine training to acquire the sound/location of pellet delivery, with single pellets dispensed with an intertrial interval (ITI) of 100±40s throughout a 30-minute session. Then, rats underwent lever press training, learning to depress a single lever for pellet delivery on a fixed ratio-1 schedule. Rats trained on the right or left lever in separate sessions, with each phase culminating when rats performed > 30 presses in a session. In some cases, multiple shaping procedures were run daily to expedite training.

Rats were next trained in nosepoke shaping, in which they learned to poke into a lit food trough to elicit a reward-associated lever. The trough was illuminated for 10 seconds, then a poke caused extension of either the right or left lever along with extinguishment of the trough light. A press on the lever evoked a single pellet delivery as well as immediate lever retraction. After a 10±2 ITI, the next trial began with the opposite lever extended after the nosepoke. After >60 successful trials, rats progressed to reward magnitude discrimination, during which they learned to discriminate between levers yielding small vs large rewards. This was comparable to nosepoke shaping with two main differences: first, both levers were simultaneously extended, providing rats with a choice between two options. Second, one lever produced a single pellet delivery (small reward), whereas the other delivered 3 pellets (large reward). Each session consisted of 50 trials, and rats trained until demonstrating at least 70% choice of the large reward option.

Risky Decision-making Task (RDT)

Detailed RDT methodology was described in previous work (Orsini and Simon, 2020). In summary, rats were trained to choose between levers associated with a small reward (1 pellet) or a large, risky reward (3 pellets and escalating risk of 1-sec foot shock). Procedure was comparable to reward magnitude training: each trial began with food trough illumination, and a nosepoke into the lit trough within 10 seconds caused both levers to extend. After rats chose between the small and large/risky options, both levers were retracted, and the reinforcer/punishment was delivered immediately. After the outcome, the trial proceeded to a 10± second ITI followed by the next trial. If the subject failed to make a choice within 10 seconds, the trial was scored as an omission and proceeded to the ITI.

Each session consisted of 5 blocks of 18 trials, with each block containing a different risk of shock (0,25,50,75,100%). The first 8 trials were “forced choice” with only a single lever extended after each trough nosepoke (four trials of each lever presented in pseudorandom order). The next 10 were “free choice”, with both levers extended after each nosepoke and rats selecting between options. Shock was set at an intensity of .2mA throughout adolescence and during the first RDT assessment in adulthood, then shifted to .3mA for a second assessment in adulthood. To facilitate completion of RDT during the brief adolescent window, all adolescent rats trained seven days/week.

Shock Threshold Testing

To determine whether MSEP caused an overall increase in shock tolerance, a subset of rats (n = 15, 10 MSEP/5 control) were given shock threshold testing (Jacobs and Moghaddam, 2020; Minnes et al., 2023). In brief, rats were placed into an operant chamber with a shock-generating floor, then foot shocks were presented for 5-s intervals with a 10-s interval between each trial. Shock intensity began at 0.05 mA and was increased 0.02 mA with each additional presentation. The rats reached threshold after the shock evoked movement of all four limbs or vocalization, with a higher shock threshold indicative of greater pain tolerance.

Adolescent Training Modifications

To confirm that RDT training was complete during the brief adolescent window, subjects trained seven instead of the standard five days/week. When subjects were slower to acquire lever press or nosepoke shaping, they performed two shaping sessions daily, with a minimum of one hour between sessions. Finally, during magnitude discrimination training, rats normally do not proceed to RDT until displaying >75% preference for the large reward along with completing > 50% of trials (Gabriel et al., 2019, 2023; Liley et al., 2022). Due to limited training time and increased satiation of the adolescent rats, subjects progressed to RDT after >75% large reward choice even with >50% trial omissions.

Statistical Analysis

Risky decision-making was reported as % risky choice within each individual block ((# of risky choices in block / total choices in block) x 100). This was then averaged across the last 5 days of training. To confirm that effects of maternal separation were not solely driven by this metric, we also compared a raw measure of total risky choices between MSEP and control groups. Stable performance across these sessions was confirmed by lack of effect of session and lack of interaction in a session X block mixed ANOVA. Choice latency was defined as mean time required to choose either the risky or safe lever across all trials in the final four task blocks (because no risk was associated with either option in block one). Additionally, we compared latency to choose the large, non-risky reward in block 1 with choice of the large, risky reward averaged across blocks 2-5. Omissions were scored as total uninitiated (failure to nose poke for lever extension) and incomplete (failure to select a lever) trials. Effects of maternal separation on each variable was measured with a treatment (maternal separation vs control) X sex X block mixed ANOVA. Age effects were assessed using age (adolescent vs adult) X treatment X sex X block mixed ANOVA. Finally, shock threshold was analyzed with a sex X treatment ANOVA. While all significant effects are reported, we omitted some non-significant interactions to keep the results concise.

Results

Maternal Separation increases Risky Decision-making in Adolescence

Male and female rats exposed to maternal separation (MSEP) or regular conditions began training for RDT in early adolescence (PD25). We first compared the rate of RDT shaping between groups, and observed that MSEP rats completed lever press shaping more rapidly than controls (MSEP mean: 4.00 sessions, Control mean: 6.28 sessions; t(34)=6.177, p<.001). There were no group differences in nosepoke shaping (t(34)=1.996, p=.054) or magnitude discrimination (t(34)=.151, p=.881). In total, MSEP rats completed shaping more rapidly than control rats ((t(34)=2.524, p=.016). In addition, we compared total large reward choice during the final session of magnitude discrimination training, and found no difference between groups (F(1,32)=.634, p=.432), suggesting that groups displayed comparable preference for the large reward option prior to the addition of risk.

Next, we assessed risky decision-making in RDT using a sex X treatment X block ANOVA. As seen previously (Gabriel et al., 2019; Orsini et al., 2019; Farrell et al., 2021), rats shifted preference away from the large, risky reward as risk increased (main effect of block: F_(4,120)=21.615, p<.001; Fig 2a). Importantly, MSEP caused a significant increase in choice of the risky option (main effect of treatment: F_(1,30)=12.220, p=.001 [percent risky choice]; F_(1,30)=14.048, p<.001 [total risky choice]; fig 2a). There was also a significant treatment X block interaction (F_(4,120)=4.405, p=.002), such that MSEP rats displayed a reduced shift away from the large reward compared to control rats (individual sex X block ANOVAS: MSEP: F_(4,68)=8.117, p<.001; control: F_(4,60)=14.377, p<.001). There was no effect of sex on decision-making (sex X treatment X block ANOVA, main effect of sex: F_(1,30))=1.468, p=.235) and no sex-based interactions (ps>.585; Figure 2b–c).

Figure 2. Effects of maternal separation on risky decision-making in adolescence.

A. MSEP increased preference for risky options in adolescence. This was comparable between female (B.) and male rats (C.). D. MSEP reduced latency to make both risky and safe choices; dividing male and females revealed that this effect was only significant in male rats (E-F). All data in A-C depicted as mean±SEM; D-F show mean and individual scores. * depicts (A-C) main effect of treatment from sex X treatment X block ANOVA, (D-F) main effect of treatment from sex X treatment X lever ANOVA.

We next assessed response latency, defined as average time taken to choose either the safe or risky option. Using a sex X treatment X lever (risky vs safe) ANOVA, we found no difference in latency to choose the risky vs safe outcome (F_(1,22)=.785, p=.385). There was a main effect of treatment (F_(1,22)=4.773, p=.040), such that adolescent rats exposed to MSEP made choices more quickly than adolescent controls (Figure 2d). There was no lever X treatment interaction (F(1,22)=.005, p=.943), indicating that the reduction in response latency in MSEP rats was comparable across risky and safe choices. There was no effect of sex ((F(1,22)=2.079, p=.163), although there was a significant sex X treatment interaction (F(1,22)=5.227, p=.032) and near significant lever X treatment X sex interaction, (F(1,22)=3,164, p=.089). We analyzed this further with individual treatment X lever ANOVAs for each sex, which revealed that females exposed to MSEP showed no differences in latency compared to controls (p=.944), whereas MSEP males showed reduced latency to select either option (p=.007; Figure 2d–f).

To determine the effects of risk on large reward choice latency, we used a sex X treatment X block ANOVA to compare latency to choose the large reward in block 1 (no risk) vs. the other blocks (25,50,75, & 100% shock). We found a main effect of block such that latency was increased when risk was added to the large reward (risk-free mean: 2.35s, risk mean: 3.50s; F(1,30)=54.443, p<.001). Interestingly, there was a significant effect of treatment (F(1,30)=7.214, p=.012) and treatment X block interaction (F(1,30)=4.270, p=.048), such that groups selected the large, safe reward at a comparable latency (MSEP mean = 2.29s, control mean = 2.42s), but MSEPs selected the risky option more quickly than controls (MSEP mean = 3.14s, control mean = 3.90s).

Finally, we assessed omissions, defined as trials in which rats failed to initiate or complete a trial, with a sex X treatment X block ANOVA (Table 1). There was a significant effect of block (F(4,128)=190.295, p<.001), such that subjects on average omitted more trials with increasing risk. There was also an effect of treatment (F(1,32)=8.454, p=.007), with adolescent rats exposed to MSEP omitting less than controls (MSEP: mean=4.92/block, SEM=.26, Control: mean=6.24/block, SEM=.38). There was a significant effect of sex (F(1,32) = 4.947, p=.033) and block X sex interaction (F(4,128)=3.123, p=.017), with females omitting more trials than males (Figure 2e–f).

Table 1. Risky decision-making Task Omissions.

All data represent total omissions/block and are depicted as mean±SEM.

	Maternal Separation					Control
Risk Block	0%	25%	50%	75%	100%	0%	25%	50%	75%	100%
Adolescent
Female	2.00±.35	3.18±.42	5.74±.36	6.64±.44	8.44±.48	3.26±.40	5.66±.60	7.46±.45	8.76±.33	9.00±.28
Male	2.13±.35	2.50±.63	3.00±.69	5.97±.57	8.38±.33	1.65±35	4.33±1.04	5.8±1.10	6.73±.88	9.00±.37
Adult (.2mA)
Female	3.26±.37	4.86±.58	6.96±.78	7.48±.81	8.48±.96	4.12±.58	4.76±.61	7.06±.78	7.22±1.01	7.64±.71
Male	0.88±.46	1.08±.64	1.85±1.07	1.88±.83	2.30±1.02	1.43±.37	2.03±.842	2.60±.90	3.34±1.01	5.43±.93
Adult (.3mA)
Female	3.22±.47	6.34±.79	7.56±.72	8.66±.77	8.86±.53	4.56±.73	6.02±.97	6.92±1.02	7.86±.90	8.56±.88
Male	1.28±.59	3.40±1.13	4.05±1.29	5.28±1.33	5.98±1.33	3.41±1.13	2.00±.98	2.52±1.12	4.30±1.01	4.75±.97

Maternal Separation does not affect Risky Decision-making in adulthood

After a 10-day break, rats resumed RDT performance with the same parameters as in adolescence, and risk-taking was again measured and compared between groups on PD85 with a sex X treatment X block ANOVA. As previously, there was a main effect of block, with reduced choice of the large reward as risk of punishment escalated (F(4,120)=16.02, p<.001). Unlike during adolescence, there was no main effect of treatment on risky decision-making (percent risky choice: F(1,30)=.255, p=.617; total risky choice: F_(1,30)=1.572, p=.219) or treatment X block interaction (F(4,120)=1.956, p=.106; Figure 3a). Also in contrast with adolescence, there was an effect of sex (F(1,30)=28.564, p<.001) and sex X block interaction (F(4,120)=5.657, p<.001), such than females showed reduced preference for large, risky rewards compared to males (Figure 3b–c). There were no significant sex X treatment X block or sex X treatment interactions (ps>.250), suggesting that the sex differences in risk-taking manifest in adulthood regardless of exposure to early life MSEP.

Figure 3. Effects of maternal separation on risky decision-making in adulthood.

A. MSEP did not affect preference for risky options in adult female (B.) or male rats (C.). D-F. MSEP did not significantly affect latency to make risky or safe choices. All data in A-C depicted as mean±SEM; D-F show mean and individual scores. All data depicted as mean±SEM.

We next compared choice latency between MSEP and control adults using a sex X treatment X lever ANOVA. There was no difference in latency to choose safe vs risky options in adulthood (main effect of lever: F(1,26)=2.160, p=.154; Figure 3a). There was a non-significant trend toward MSEP reducing latency to select either option (F(1,26)=4.423, p=.091), but no lever X treatment interaction (F(1,26)=.079, p=.781). Unlike in adolescence, there was a significant main effect of sex on latency (F(1,26)=15.731, p<.001), such that males made decisions more quickly than females (Figure 3e–f). There were no significant sex-based interactions, suggesting that both lever type and early life experience did not influence sex differences in response latency (ps>.469). We next compared latency to choose the large reward when risk-free (block 1) with risky large reward choice (averaged blocks 2-4) using a sex X treatment X block ANOVA. As with adolescents, there was a main effect of block, with risky reward choice taking longer than safe choice ((F(1,30)=39.609, p<.001). However, there was neither a significant effect of treatment (although it approached significance; F(1,30)=39.5, p=.056) nor a significant treatment X block interaction (F(1,30)=.252, p=.619).

We next analyzed omissions with a sex X treatment X block ANOVA (Table 1). As previously, there was an effect of block (F(4,128)=38.913, p<.001), p<.001), such that all subjects omitted more trials as risk of punishment increased. There was no effect of treatment on omissions (F(1,32)=.952,p=.337; MSEP: mean=4.65/block, SEM=.82, Control: mean=5.19/block, SEM=.66). As in adolescence, females omitted more trials than males (F(1,32)=33.196, p<.001), but there was no sex X treatment interaction (F(1,32)=1.095, p=.303).

Retesting RDT in Adults with Increased Shock Intensity

During adult RDT, male preference for risky rewards approached ceiling levels regardless of treatment (Figure 3c). It is possible that effects of MSEP in adulthood were concealed by this near-maximal choice of the risky option. To create more parametric space for detecting potential group differences, rats were retrained in RDT with shock intensity raised from .2 to .3 mA, which has been shown to reduce risky choice (Shimp et al., 2015). As anticipated, increasing the shock significantly reduced risky choice (shock intensity X block ANOVA: F(1,30)=53.756, p<.001). When isolating the .3 mA condition and analyzing data with a sex X treatment X block ANOVA, there was a significant main effect of block (F(4,120)=66.567, p<.001), with subjects shifting preference away from large rewards with increasing risk. As with the lower shock intensity, there was no effect of treatment on risky decision-making (% risky choice: F(1,30)=.095, p=.760; total risky choice: F(1,30)=.009, p=.926) or risk X treatment interaction (F(1,30)=.012, p=.915; Figure 4a). There was an effect of sex, such that males chose risky options more than females (F(1,30)=8.094, p=.008; Figure 4b–c). There were no other sex-based interactions (ps>.215). Therefore, MSEP had no effect on adult risky decision-making regardless of shock intensity.

Figure 4. Effects of maternal separation on risky decision-making in adult rats with increased punishment intensity.

A. MSEP did not affect preference for risky options in adult female (B.) or male rats (C.). D. MSEP reduced latency to make both risky and safe choices. This group distinction was comparable across both females (E.) and males (F.). All data in A-C depicted as mean±SEM; D-F show mean and individual scores. * depicts main effect of treatment from sex X treatment X lever ANOVA.

There was no significant difference in latency to choose the safe vs risky options (sex X treatment X lever ANOVA: F(1,27)=.621, p=.437). However, there was a significant difference between treatment groups (F(1,27)=6.225, p=.019), with MSEP reducing latency to choose either option (Figure 4d). There was no main effect of sex (F(1,27)=.218, p=.645), and no significant interactions between variables (ps>.653; Figure 4e–f). As in other conditions, trial omissions were elevated as risk of punishment increased (sex X treatment X block ANOVA: F(4,28)=65.768,p<.001; Table 1). There was no effect of treatment on omissions (F(1,32)=.578,p=.453; MSEP: mean=6.44/block, SEM=.74, Control: mean=5.58/block, SEM=.80). Females omitted significantly more trials than males (F(1,32)=16.941, p<.001), with no sex X treatment interaction (F(1,32)=.345, p=.561).

Finally, a subset of rats performed shock threshold testing to determine if shock/pain tolerance was altered by MSEP in adulthood. A sex X treatment ANOVA revealed that MSEP rats had a higher shock threshold than controls, indicative of higher pain tolerance (MSEP shock threshold mean: .445mA, control mean: .230mA; F(1,11)=59.92, p<.001)). There was also an effect of sex, with males displaying elevated shock threshold compared to females (male mean: .378mA, female mean: .288mA ;F(1,11)=9.839, p=.009). There was no sex X group interaction, suggesting that sex differences in shock sensitivity were not influenced by maternal separation (F(1,11)=.001, p=.978.).

Comparing Risky Decision-making in Adolescence and Adulthood

We utilized an age X sex X treatment X block ANOVA to longitudinally compare RDT across adolescence and adulthood. There was a near-significant trend of age (F(1,20)=3.227, p=.083), with adolescents demonstrating increased risk preference compared to adults. There was also a significant main effect of sex (F(1,20)=16.605, p<.001), with males showing increased risk-taking across adulthood/adolescence. Critically, there was a significant age X treatment interaction (F(1,120)=11.502, p=.002), enabling individual comparisons between ages within both MSEP and control subjects (Figure 5). A sex X age X block ANOVA limited to MSEP rats revealed that this group was significantly more risk-prone during adolescence than adulthood (main effect if age: F(1,16)=11.718, p=.003). There was also a significant sex X age interaction (F(1,16)=16.574, p<.001) such that MSEP males displayed consistent risk preference across the lifespan, whereas females were riskier during adolescence than adulthood. Conversely, no effects of age on risk-taking were observed with a comparable ANOVA limited to controls (F(1,14)=2.898, p=.111). In summary, there was no difference in risky decision-making between adolescence and adulthood in subjects raised in normal conditions, but rats exposed to MSEP during development were riskier in adolescence than adulthood.

Figure 5. Longitudinal comparison of risky decision-making across the lifespan.

A. MSEP rats showed increased risk-taking in adolescence compared to adulthood. This difference was not observed in control rats. B. During adolescence, male and female rats displayed comparable preference for risky options. In adulthood, males were more prone to risky choice than females. Each data point represents mean choice of the risky option across the entire session for an individual rat, with group means visualized as a line. *depicts main effect of age from age X sex X treatment block ANOVA (A) and main effect of sex from sex X treatment X block ANOVA in adults only.

Discussion

We determined that MSEP increases risky decision-making in both male and female rats during late adolescence (post-natal days 55-60). Additionally, adolescents exposed to MSEP made both risky and safe decisions more rapidly than controls. In adulthood, MSEP-evoked risk-taking and increased rate of decision-making were both abolished. MSEP and control adults also showed comparable risk-taking with increased shock intensity, although this was sufficient to reinstate the increased rate of choice in MSEP rats. Interestingly, no sex differences in risk-taking were observed during adolescence, then males shifted toward a higher risk decision-making strategy than females in adulthood. Finally, MSEP-exposed rats showed a sex-dependent reduction in risk preference from adolescence to adulthood, whereas there was no age difference in risk-taking in control rats.

Maternal Separation Increases Risk-taking in Adolescence

Adolescence is characterized by extensive behavioral and neurobiological alterations, making the brain and behavior particularly vulnerable to adversity. We observed that in late adolescence (days 55-60), MSEP increased risky choice in RDT. Moreover, MSEP-evoked risk-taking was comparable between male and female rats. This was somewhat surprising, as the consequences of MSEP are often sex-specific (Spivey et al., 2009; Kunzler et al., 2015; Ganguly et al., 2019; Honeycutt et al., 2020; Zühlsdorff et al., 2022; Rincón-Cortés, 2023; Gildawie et al., 2024). This suggests that the MSEP effects on adolescent risk-taking are likely not mediated by alterations in sex hormone transmission or other sexually differentiated neural mechanisms.

Brain regions involved with regulation of risk-taking, such as prefrontal cortex and striatum (Simon et al., 2011; Orsini et al., 2015a; Freels et al., 2020; Truckenbrod et al., 2023a), are still undergoing development in adolescence (Geier et al., 2010; Simon and Moghaddam, 2015; Galvan and Tottenham, 2016). In addition, punishment-based risk-taking is modulated by dopamine transmission, and the dopamine system is still undergoing development during adolescence (Andersen et al., 2001; Ernst et al., 2009; Walker et al., 2010; McCutcheon et al., 2012; Matthews et al., 2013; Simon and Moghaddam, 2015; Kim et al., 2016). It is possibly that ongoing development renders these brain regions particularly vulnerable to the effects of MSEP, resulting in enhanced risk-taking. Mesolimbic dopamine release following an acute stressor is enhanced in adolescents who endured MSEP (Jahng et al., 2010); this hyperactive dopamine response is analogous to the increased phasic dopamine release in rats with a bias toward risk-taking (Freels et al., 2020), and may be related to elevated risk-taking after MSEP. MSEP also alters dopamine receptor expression in prefrontal cortex and striatum in adolescent females (Majcher-Maślanka et al., 2017), another biomarker of punishment-driven risk-taking (Simon et al., 2011; Mitchell et al., 2014). MSEP causes differences in both structural and functional measures of communication between basolateral amygdala and medial prefrontal cortex (Honeycutt et al., 2020; Thomas et al., 2020), regions that contribute to different aspects of both punishment-based and reward omission-based risk-taking (Ghods-Sharifi et al., 2009; Onge et al., 2012; Orsini et al., 2017, 2018; Tremblay et al., 2021). Therefore, it is possible that increased adolescent risk-taking following MSEP is a result of aberrant communication in cortical/striatal/limbic circuitry.

MSEP reduced latency to make both risky and safe decisions during adolescence. In addition, while all rats displayed increased latency to choose risky compared to safe large reward options, this safe vs. risky choice difference was diminished in MSEP animals. Collectively, these data suggest that MSEP causes more rapid selection of risky choices. This may result from reduced processing of potential outcomes during pre-choice deliberation, leading to underestimation of risk. This reduced decision latency may be related to MSEP altering activity in basolateral amygdala (BLA), as BLA lesions increased choice of punished or risky rewards and reduced latency to perform these actions (Jean-Richard-dit-Bressel and McNally, 2015; Orsini et al., 2015b). Additionally, during pre-choice deliberation, lateral orbitofrontal cortex signals information about impending risk, and this signaling is attenuated in risk-taking subjects (Gabriel et al., 2024). It is possible that MSEP diminishes processing in either or both brain regions, leading to lack of sensitivity to punishment or miscalculation of punishment risk.

MSEP also may enhance risk-taking by reducing cognitive flexibility, hampering the ability to shift choice after changes in risk and “flattening” the decision-making curve (Figure 2). A meta-analysis determined that MSEP reduces flexibility in rodents, although these results were somewhat heterogeneous (Ou-Yang et al., 2022). MSEP reduced flexibility in post-weaning animals (day 26) but not in late adolescence/early adulthood (day 60), the time period in which risk-taking was assessed in the current study (Thomas et al., 2020). Additionally, risk-taking rats display increased cognitive flexibility compared to risk-averse subjects (Shimp et al., 2015), suggesting that impaired flexibility would likely manifest as reduced rather than enhanced risk-taking.

MSEP-enhanced risk-taking may have been driven by decreased pain sensitivity, since the punishment used in RDT is a nociceptive stimulus (mild shock). We tested shock sensitivity in adulthood, and found that MSEP animals were less sensitive to footshock than controls. This contrasts with previous reports showing that MSEP increases nociception in both adolescent and adult rodents (Amini-Khoei et al., 2017; Melchior et al., 2018; Salberg et al., 2020; Gieré et al., 2023), which would be expected to reduce risk-taking rather than the enhancement observed here. This contrast may be related to the rats in the current study experiencing extended exposure to controllable shocks prior to shock threshold testing. Additionally, statistical testing may have been affected by the imbalanced sample size (control n=5, MSEP n=10), although the effect size was substantial (η²=.719). Notably, this pain tolerance was only assessed in adulthood, where there was no difference in risk-taking between groups. Moreover, several measures of pain tolerance and shock sensitivity (including the shock threshold testing used here are uncorrelated with risk-taking in RDT (Simon et al., 2011; Truckenbrod et al., 2023b), and administration of the analgesic drug morphine does not alter risk-taking (Mitchell et al., 2011), suggesting that RDT performance is not tightly linked to pain tolerance. Therefore, it seems unlikely that MSEP-induced adolescent risk-taking is solely driven by altered pain tolerance.

The onset of substance use disorder (SUD) often occurs in adolescence, and early life neglect is associated with increased substance use during this period (Edalati and Krank, 2016; Yoon et al., 2024). Risky decision-making and insensitivity to punishment are commonly observed in SUD, and can drive ongoing drug seeking in the face of health, financial, and social consequences (Bechara et al., 2001; Brand et al., 2008; Guttman et al., 2018; Jean-Richard-Dit-Bressel et al., 2018; Chen et al., 2020; Ariesen et al., 2023). It is possible that elevated risk-taking is one of the factors that promotes SUD in adolescents with early life adversity. Risk-taking in RDT is associated with a cluster of addiction relevant traits, including impulsive action (Gabriel et al., 2019, 2023), cue sensitivity (Olshavsky et al., 2014), and sensitivity to drugs of abuse (Mitchell et al., 2014; Gabriel et al., 2019; Garman et al., 2021). In addition, risk-taking is associated with a hypersensitive mesolimbic dopamine system and reduced D2 receptor expression in dorsal striatum, both of which are observed in people with a history of substance misuse (Volkow et al., 2007; Simon et al., 2011; Freels et al., 2020). Therefore, elevated risk-taking caused by MSEP may engender a state of vulnerability to substance use in adolescence.

One caveat of early life stress protocols such as MSEP is the inability to determine if effects are limited to stress during development, or if these effects are would still occur after chronic stress delivered at other points in the lifespan. Several studies have measured effects of chronic stress in adulthood on risk-taking, with variable results. Chronic immobilization stress in adulthood increased risky decision-making (Friedman et al., 2017; Mudra Rakshasa and Tong, 2020), and adolescent isolation stress caused an increase in high-risk, sub optimal choices on the rat gambling task (Zeeb et al., 2013). Conversely, chronic unpredictable stress caused risk aversion (Morgado et al., 2015), and repeated social defeat stress had no effect on risk-taking (Boutros et al., 2017). Ultimately, there is converging evidence that risky decision-making is vulnerable to chronic stress across the lifespan, with the current report demonstrating that even very early life stress can cause increased risk-taking that persists for weeks after the stressor is terminated.

MSEP does not increase risk-taking in adulthood

While MSEP increased risk-taking in adolescence, this effect did not persist into adulthood. Moreover, the reduced latency to make a choice in adolescence was abolished in adulthood, providing evidence that MSEP-evoked adolescent risky choice was related to “rushed” risky decision-making. However, raising shock intensity in adulthood caused this group difference to re-emerge, suggesting that MSEP-evoked may indeed cause changes to choice latency that endure into adulthood, but these only manifest during higher cost risk-taking scenarios.

These data contrast with a previous study reporting increased disadvantageous choice in the rat gambling task (RGT) in adults exposed to MSEP (Cao et al., 2016). This distinction may be a result of differences in the risk of punishment used in these tasks; RDT uses a probabilistic foot shock, whereas the RGT uses risk of a time out periods in which reward is no longer available. Differences have been noted in the biological processes mediating these modalities of risk-taking (Orsini et al., 2015a), so the divergent effects of MSEP are not entirely unexpected. Additionally, MSEP has also been shown to increase impulsive action in the differential reinforcement of low rates of responding task (DRL) in adults (Lovic et al., 2011). Impulsive action in DRL is correlated with risk-taking in RDT(Gabriel et al., 2019, 2023); therefore, it was surprising that increased risk-taking did not manifest in adulthood in the current report. It is possible that early life training in RDT followed by re-testing in adulthood in the current study was sufficient to mask the risk-enhancing effects of MSEP. This remains to be tested by assessing RDT in naive rather than pre-trained MSEP-exposed adults.

It is possible that MSEP-induced adolescent risk-taking was caused by impaired task acquisition, leading to a bias toward one response option (risky) due to incomplete understanding of task contingencies. This would explain why MSEP-induced risky choice did not persist from adolescence to adulthood, as the additional training may have been sufficient for MSEP rats to achieve the same task comprehension as controls. However, evidence suggests that MSEP did not induce broad learning deficits. First, MSEP rats displayed no deficits in acquisition of task shaping protocols, including the complex magnitude discrimination training preceding RDT. On the contrary, MSEP rats displayed increased rate of shaping, suggesting that some learning processes may be accelerated rather than impaired after MSEP. Second, to support the argument that learning was not ongoing in adulthood, we analyzed risky choice on day one of adult RDT (15 days before the data reported in Figure 3). We found that MSEP and control subjects were comparable on this first day (F(1,16)=.130, p=.723; data not shown [note: degrees of freedom are low because omissions were relatively high on this session]), suggesting that MSEP performance immediately shifted to a less risk-preferring strategy in adulthood rather than shifting after weeks of additional training. Finally, rats trained in another punishment-based task, the Delayed Punishment Decision-making Task (Liley et al., 2019), one month after completing RDT. After completion of training, there was no difference between MSEP and controls in this task (F(1,20)=.552, p=.466; data not shown), providing further evidence that ability to learn a cost benefit decision-making task did not differ between groups. Collectively, these data provide strong evidence that MSEP did not impair RDT task acquisition, but instead selectively increased risk-taking in adolescence. However, running a comparable experiment wherein MSEP rats are exclusively tested during adulthood is necessary for definitive evidence of adolescent-selective MSEP-evoked risk-taking.

The reduction of risk-taking to a level comparable to controls in adulthood suggests that task-relevant brain development is affected by MSEP, but ongoing development in late adolescence/young adulthood is sufficient to overcome these alterations. Therefore, it may be beneficial for future research on mechanisms underlying MSEP-induced risk-taking to focus on adolescence, a period of vulnerability to psychiatric disorders (Simon and Moghaddam, 2015; Blakemore, 2019).

Comparing risk-taking across the lifespan

While punishment-based risky decision-making has been previously assessed in adolescent male rats (Mitchell et al., 2014), this study is the first to compare longitudinal measurements of risk-taking in adolescence and adulthood of male and female rats. We observed that in control subjects, risky decision-making was comparable between adolescence and adulthood. However, age differences emerged in rats exposed to MSEP, with MSEP adolescents showing greater risk-taking than adults. This suggests that increased adolescent vs adult risk-taking may not be inherent, but instead manifests after early life adversity.

Sex differences in RDT have been reported repeatedly, with females adopting a more risk-averse strategy than males (Orsini et al., 2016, 2021; Blaes et al., 2022; Gabriel et al., 2023; Truckenbrod et al., 2023b). Interestingly, there were no sex differences during adolescence regardless of early life experience. Sex differences then manifested in early adulthood, with the first adult measurement occurring only 25 days after the adolescent measurement (days 56-60 vs days 81-85). Therefore, functional sex differences within brain circuitry underlying risk and reward during decision-making may not fully emerge until adulthood.

Conclusions

These data reveal that MSEP increases risky decision-making, but this shift is limited to adolescence. This increased adolescent risk-taking may contribute to the long-term adverse outcomes of early life adversity by promoting ongoing reward seeking in the face of negative consequences. Future research will aim to clarify the neuronal mechanisms underlying this altered decision-making, while also delineating whether MSEP-induced risk-taking is regulated by reduced sensitivity to punishment risk or increased sensitivity to reward.

Highlights.

• Maternal separation increased risk-taking in adolescence

• Latency to make a risky choice was also reduced after maternal separation

• Enhanced risk-taking did not persist into adulthood

• Sex differences in risk-taking first manifest in adulthood

• Risk-taking is comparable between adolescence and adulthood in control subjects

Abbreviations:

ELA

early life adversity

MSEP

maternal separation

RDT

risky decision-making task

PD

postnatal day

ITI

intertrial interval

SUD

Substance Use Disorder

BLA

basolateral amygdala