Abstract

Pyramidal cells in the dorsal hippocampus (dCA1) are excitatory neurons modulated by environmental cues. While a population of dCA1 cells encodes spatial location, other groups are activated by reward probability and encounters. Since “rewards” are predicted at “locations,” we sought to determine how spatiotemporal coding patterns in the dCA1 resolve contextual preference and subsequent change in preference that is driven by reward encounters. Specifically, we examined these encoding patterns in biased place-preference tasks for simple reward acquisition and for complex discrimination of reward magnitudes. Initial behavioral tests for mice without neural implants revealed a higher sensitivity to discriminating between two locations associated with reward magnitude, in comparison to reward detection. Analysis of dCA1 single-unit spatiotemporal activity during pre-conditioning revealed that these cells exhibited peak firing as they approached the less preferred context. Therefore, when the contextual preference is biased toward a reward or a higher-magnitude reward, a change in dCA1 firing rate around context entry events reflects the updated spatial preference. Interestingly, the context of lower preference with no associated reward or a lower-value reward elicits a stronger firing response than the alternative contexts with higher reward values. Together, we conclude that the spatiotemporal firing patterns of dCA1 single units and the threshold of peak FR change encode contextual preference. Ultimately, the spatiotemporal pattern is updated (remapped) when there is a change in the contextual preference driven by reward contingencies.

Introduction

Neural circuits are central to adaptive behavioral mechanisms that govern survivability in an organism’s natural habitat (Tovote et al., 2015; Plas et al., 2024; Park et al., 2021; Zhou et al., 2017). These circuits cut across the mesocorticolimbic and forebrain centers and underlie innate defensive behavior (Zhao et al., 2023; Carli and Farabollini, 2022), motivated feeding/foraging (Sternson et al., 2013; Hao et al., 2020), stress response/fear, reproductive behavior, and context (valence) discrimination (Adams et al., 2012; Fischer and O’Connell, 2017; Rapp and Nawrot, 2020; Kassraian et al., 2023). Reciprocal connections between the mesocorticolimbic and forebrain circuits detect environmental cues about reward and aversion, and facilitate their recall to guide future decision-making (Cui et al., 2023; Castillo Díaz et al., 2022; Dalenberg et al., 2018; Coley et al., 2021). Adaptive learning is evident in place preference/aversion tasks, where animals show an affinity for locations, cues, or events generally associated with a reward and tend to avoid aversive places or cues (Napier et al., 2013; Roux et al., 2003). To recapitulate complex environmental scenarios, previous studies have shown that animals discriminate between cues or locations based on a mixed probability of getting rewarded with punishment (Anderson, 2019; Orsini and Simon, 2020).

The hippocampus encodes information about space, location, and context. The dorsal hippocampus (dCA1) plays a crucial role in working memory and facilitates the computation of novelty for long-term storage (Dimsdale-Zucker et al., 2022; Liu et al., 2022; McKenzie and Buzsaki, 2016; Russo et al., 2024). The role of the hippocampus in novelty detection stems from its mechanism of comparing new information entries in working memory with previously stored cortical memory (Lisman, 1989, 1999; Lisman and Grace, 2005). Since the environment is enriched with novel sensory stimuli, weight and valence assignments are bolstered by contextual and temporal relevance (Barbeau et al., 2017; Frank and Kafkas, 2021; Gómez-Ocádiz et al., 2022; Valenti et al., 2018). Novel stimuli that directly relate to survival, adaptive behavior, rewards, and punishment are prioritized for long cortical storage and guide future behavioral response (Braun et al., 2018; Cowan et al., 2024; Kalbe and Schwabe, 2022).

In addition to the well-studied spatial coding place cells, a population of dCA1 neurons that predict or detect rewards along the exploration path has been described (Gauthier and Tank, 2018; Jarzebowski et al., 2022). Since spatial novelty can be rewarding, there is a rationale to test whether spatiotemporal mapping of dCA1 reward-sensing ensembles can predict spatial preference and a change in spatial preference that is driven by reward contingencies. Based on this concept, the current study proposes that the spatiotemporal maps of dCA1 ensembles can predict contextual behavioral preference, and that these maps are updated to reflect a change in the preferred context. To test this concept, a sucrose conditioned place preference (sCPP) task was modified such that the initial preconditioning phase depicts the animal’s initial spatial preference, and subsequent sucrose reward bias was aimed at driving a change in the spatial preference. Therefore, sampling spatiotemporal peak firing rates of dCA1 ensembles during the preconditioning and test phases provides a premise for assessing ensemble dynamics relative to spatial preference and a change in preference driven by an absolute or a magnitude reward comparison.

Our results showed that dCA1 neurons were more responsive to reward probabilities than reward encounters. For these tasks, entry events into less preferred contexts, reward omissions, or rewards of lower value elicited stronger responses than preferred context entry, reward encounters, or encounters of reward with a higher value. The spatiotemporal firing pattern of dCA1 putative units encodes the initial pre-conditioning contextual preference and reflects a preference change when the sCPP task is biased with reward modalities. Spatiotemporal dCA1 putative unit population dynamics also showed that magnitude discrimination tasks produce stronger memory traces than simple rewards detection tasks. As such, when rewards were omitted during the test sessions, the change in peak FR for dCA1 neurons was significantly greater than the change during reward encounters in the conditioning phase. Conversely, for reward-detection tasks, the change in firing rate was comparable during conditioning reward encounters and during test reward omission.

Materials and methodsExperimental animals

Two separate groups of mice were used for the experiments. In the first cohort (n = 20), mice without neural implants were used to assess the behavioral response to absolute and magnitude sCPP task modalities. In the second cohort (n = 7), mice had neural implants and were used to evaluate spatiotemporal coding during sCPP tasks. The IACUC committee at Louisiana State University approved all procedures. Mice were kept in standard laboratory housing with alternating light/dark cycles (12:12 h). Food and water were provided ad libitum. Mice without implants were housed in groups of 5 per cage (same sex). Animals with neural electrode implants were housed individually and allowed to socialize in controlled laboratory settings.

Sucrose conditioned place preference (sCPP)

Adult C57BL/6 mice (n = 20 mice, 10M/10F, 12–16 weeks) weighing 25–30 g (no neural implant) were used for sCPP behavioral tests. A biased sucrose CPP test was used for this experiment. A shuttle box with dimensions 40 cm (L), 10 cm (W), and 12 cm (D) was divided into three sections. The central area was the starting zone, separated from the two edge compartments by removable dividers. The two edge compartments were used as target contexts and were distinguishable by three sensory (visual, tactile, and olfactory) cues (Figure 1A). Each session was 10 min and was performed on the same day for all mice.

Biased sucrose conditioned place preference (sCPP). (A) Schematic illustration of a shuttle box designed for sCPP tasks. Contexts were distinguished spatially (left/right) and by three sensory cues. (i) visually distinct wall patterns, (ii) grid or horizontal floor bars to create distinguishable tactile patterns, and (iii) light vinegar spray in one context for odor discrimination. (B) Experimental timeline for habituation (3 sessions), preconditioning (1 session), conditioning (3 sessions), and test (1 session) phases. (C) Representative picture of a mouse in a CPP test chamber showing visual/tactile cues and location of sucrose rewards pots. (D) Representative preconditioning heat maps for two mice with context A or context B preference, respectively. (E) Schematic illustration of biased sCPP task execution for mice with preconditioning context A or B preference. (F) Schematic illustration of biased reward conditioning for absolute (0%:16% sucrose) and magnitude (2%:20% sucrose) sCPP tasks.

Absolute sCPP

This is a reward detection task in which one context had a reward (sucrose) during the conditioning phase. For this task, context A had vertical lines on all three walls, no odorant, and a grid mesh floor. Conversely, Context B had horizontal lines on all walls, 10% vinegar sprayed lightly, and horizontal bar floors.

Magnitude sCPP

This is a discrimination task in which mice explored two rewards of different magnitudes during the conditioning phase. Here, the context was changed for novelty with a different mix of floor (tactile), wall (visual), and vinegar (olfactory) locations.

Mice were habituated to the testing chamber at 72, 48, and 24 h (H1–H3). Subsequently, mice were assessed in a preconditioning task to determine the spatially preferred and non-preferred contexts per mouse (Figures 1B,C). This output was determined by the comparison of the total time spent in each context during the preconditioning session. Once the initial context preference was determined per mouse, three biased conditioning sessions were performed (daily) such that a reward (16% sucrose, abs sCPP) or a reward of higher value (20% sucrose, mag sCPP) was dispensed at the preconditioning non-preferred context. Before daily conditioning sessions, food restriction was performed for 8–10 h (Figure 1B). Based on the preconditioning output, the non-preferred context was set as “target,” and the preferred context was set as “non-target” for the subsequent conditioning and test phases. Representative heatmaps of preconditioning outcomes for two mice, in which one mouse preferred context A and the other showed a preference for context B are shown in Figure 1D. Contextual preference during the preconditioning phase was determined by a greater than 55%-time distribution in one of the two contexts. For contexts A or B, PrC preference, a biased reward, or reward magnitude presentation was performed during the conditioning phase (Figure 1E). For absolute sCPP conditioning sessions (C1–C3), 16% sucrose was dispensed into the reward pot in the target context (PrC, non-preferred), and no reward was delivered in the non-target context (PrC, preferred) (Figure 1F). For mag sCPP conditioning (C1–C3), both contexts contained a reward. However, the target context (PrC, non-preferred) contained 20% sucrose while the non-target context (PrC, preferred) contained 2% sucrose (Figure 1F). Subsequently, in the test session, animals were tested without a reward present. The sCPP score was calculated as the percentage of time spent on the target side relative to the total time spent on both sides.

Neural electrode implant and spatiotemporal analysis of dCA1 ensembles

Adult male Vglut2Cre mice (n = 7) weighing 25–30 g were used for this study. Animals were anesthetized with ketamine/xylazine (100/10 mg/kg). Once the plane of anesthesia was established, a toe pinch was performed to verify the absence of sensation. The head was fixed on a stereotaxic frame. Sterile surgical preparation techniques were used. These included shaving and cleaning the scalp area with iterations of iodine and alcohol solutions. A local anesthetic agent was applied before an incision was made to expose the skull surface. By locating the bregma, anterior-posterior (AP: −2.18 mm) and medial-lateral (ML: 1.0–2.0 mm) coordinates that corresponded to the dorsal CA1 were determined and marked. A drill tool was used to prepare a 2.5 mm craniotomy over this point to expose the dura. A sharp, bent, clean needle was then used to remove the dura. Sterile saline was applied over the craniotomy to prevent dryness. A second craniotomy housed a screw to which the ground and reference wires were connected. Once the ground and reference wires were in place, 16-channel silicon probes (Neuronexus, United States), were gently lowered into the dCA1 such that electrode contact sites were in the cell body layer (DV: 1.3–1.5 mm) (Figure 2A). The probe was tethered to a head stage to detect the spontaneous activity of putative pyramidal cells (0.3–5 kHz). The anatomical location of the probe was micro-adjusted to detect robust signals on electrode contact sites in the DV range. Once the final location was set, dental acrylic was applied to cast the implant onto the skull. Seven days after surgery (recovery), animals began behavioral testing (Figure 2B). The impedance of electrode contact sites measured between 0.8 and 2.0 MΩ. The location of the neural electrodes in the dCA1 was validated post hoc using histology (Figure 2C).

sCPP tasks with dCA1 single unit sampling. (A) Schematic illustration of a neural probe implant (16-channel electrode, 4 × 4) in the dCA1. (B) Representative picture showing a mouse with neural implants and a tether in a shuttle box during an sCPP task. (C) Fluorescence (DAPI) images showing neural electrode implant sites (scale bar 1 mm). (D) Timeline for neural recording and sCPP task sessions in mice with neural implants. Note that mice with neural implants performed four conditioning sessions. (E) Schematic depicting spatiotemporal mapping of dCA1 pyramidal single unit activity relative to the mouse XY position (cm) in the shuttle box. (F) Edge histograms and cluster plots of putative dCA1 pyramidal cells based on waveform valley-to-peak duration and peak width at half height. (G) Representative waveforms and autocorrelogram of putative dCA1 pyramidal cells. (H) Representative DAPI fluorescence image showing tissue scar of recovered neural electrode shanks in the dCA1 (scale bar 0.5 mm).

Synchronized dCA1 recording and CPP task

Dorsal CA1 spikes were sampled during the preconditioning (PrC), conditioning (C1, C2, C3, and C4), and test (Tst) phases (Figure 2D). Implanted neural electrodes were tethered to a preamplifier head stage and recording controller (Intantech, United States). Through a Noldus Mini I/O box, TTL signals triggered by zone entry and exit were transmitted to the digital-IN channels of the recording controller by wired connections (Figure 2E). All behavioral tasks were acquired with Ethovision XT17 software, which uses a machine-learning algorithm for stable rodent tracking. Body point tracking was used to determine when the animal crossed into a zone. Behavior-driven TTL signals were used in time-locking zone entry events with continuously sampled dCA1 spikes.

Spike sorting and single-unit detection

In an Offline spike sorter (OFSS, Plexon, United States), extracellular spikes were pre-processed with a Butterworth filter (300–5,000 Hz) to remove anatomical drifts and local field potential artifacts. Single-unit clustering of the recorded spikes was performed by principal component analysis (PCA) to detect putative dCA1 principal neurons. An amplitude discrimination step was implemented in the OFSS to improve the signal-to-noise ratio. A lower peak crossing threshold that is 5x the root mean square (RMS) was set for each electrode channel to eliminate noise and artifact spikes (Quiroga, 2012; Swindale et al., 2017; Chung et al., 2017). Unsorted spikes were manually invalidated where necessary in OFSS. Since more than one putative unit was detected on each electrode channel, spikes were further discriminated against based on the interspike interval (Chung et al., 2017; Quiroga, 2012).

Characterizing putative pyramidal units

Waveforms of clustered units were inspected across all channels. Viable units were accepted based on the autocorrelogram (ACG), firing rate (FR, Hz), and waveform valley-to-peak time (μsec) (Figure 2F). Putative units with a valley-to-peak time range of 400–800 μs and distinct ACG peaks at 0 ms, followed by a rapid decay (50 ms), were characterized as putative dCA1 pyramidal cells (Sotres-Bayon et al., 2012; Barthó et al., 2004). Interneuron ACGs have a distinct trough at 0 ms, with sustained activity. Interneurons were further distinguished from pyramidal cells by the absence of complex spiking and higher firing rate scores and were excluded from subsequent analysis. Putative units identified as principal cells had a mean firing rate of <5 Hz (Figure 2G). Based on the shape of the waveform, ACG, and firing rate, we expect that most of these putative neurons are glutamatergic cells (Moorman and Aston-Jones, 2015; Barthó et al., 2004).

Histology and verification of dCA1 recording sites

After CPP behavioral tasks with neural recording, mice were euthanized by isoflurane exposure in a desiccator. The concentration of isoflurane was increased until the animal became unconscious, and death was confirmed by the absence of breathing and toe pinch response. Transcardial perfusion with 4% paraformaldehyde (PFA) in PBS was performed subsequently. The brain was preserved in 4% PFA-PBS and was cryopreserved in the same (fresh) solution containing 30% sucrose. A total of 50 μm cryostat sections containing the dorsal hippocampus were mounted on slides and stained with DAPI to visualize the dCA1 tissue scaring site that indicates the anatomical location of the recording electrode shanks (Figure 2H). Fluorescence images were obtained using a Ni-U fluorescence upright microscope (Nikon AR software) equipped with a Moments camera (Teledyne Canada).

Statistical analysisSucrose conditioned place preference tests (n = 20 mice, 10 M/10 F)

Statistical analysis was performed using Origin Pro 2024 software (Origin Labs, United States). Biased sCPP scores were compared for the preconditioning and test phases using a paired T-test for males, females, and all mice (male and female) (n = 20). Outcomes for the preconditioning (PrC), conditioning (C1, C2, C3), and test (Tst) sessions were compared using One-Way ANOVA (repeated measures) with Dunnett’s post-hoc test. Comparison of absolute and magnitude sCPP outcomes was done using a paired T-test.

Sucrose conditioned place preference task with dCA1 recording (abs sCPP: n = 7, mag sCPP: n = 6)

Sucrose conditioned place preference outcomes for mice with neural implants were compared across experimental sessions (PrC, C1, C2, C3, C4, Tst). Sorted spikes for sampled putative dCA1 principal cells were exported to Neuroexplorer (Nex Technologies, United States) for further analyses. Results from spike analysis and Ethovision XT17 behavioral tracking were further analyzed in OriginPro 2024 software. Normality distribution was determined using the Kolmogorov-Smirnov test. Comparison of dCA1 ensemble activity across experimental sessions was determined using One-Way ANOVA (repeated measures) with Dunnett’s post-hoc test. Change in firing rate of dCA1 neurons during context entry events in the pre-conditioning and test phases was compared using Student’s T-test.

ResultsAbsolute sCPP (n = 20, 10 M/10 F, no neural implant)

For biased sCPP tasks, the non-preferred (nP) side during preconditioning (PrC) was set as the target (Tg), and the preferred side (P) was the non-target (nTg). In the following three conditioning sessions, mice obtained 16% sucrose solution in the target (i.e., PrC non-preferred) context, and 0% sucrose in the non-target (i.e., PrC preferred) context (Figure 3A). This test is self-executing, and mice were not exposed to any reward or pre-trained during the habituation/PrC sessions. Here, the target and non-target contexts were determined for each mouse during the preconditioning phase, and the output was applied subsequently to the conditioning (learning) sessions (Figure 1E). During the test phase, no reward was presented (retrieval) in both contexts. Analysis of the sCPP% score showed that female mice had robust sensitivity to the absolute sCPP modality, such that target context exploration increased across the conditioning and test sessions. As such, the test sCPP% score for the female cohort was significantly higher than the preconditioning score (female:Figure 3B| t = −3.47, DF = 9, p = 0.007). In contrast, male mice were less sensitive to the absolute sCPP with no significant change in preconditioning and test sCPP% scores (male:Figure 3B| t = −1.27, DF = 9, p = 0.24). Combining the male and female cohorts showed that the group (i.e., M/F) was sensitive to reward learning and context discrimination in the absolute sCPP paradigm (0%/16%). This was mostly driven by female scores (all:Figure 3B| t = −3.043, DF = 19, p = 0.007).

sCPP% score for absolute sCPP conditioned with 0%:16% sucrose. (A) Schematic illustration of absolute(abs) sCPP with biased reward conditioning. (B) Composite graph comparing preconditioning and test sCPP% scores for female (t = –3.47, DF = 9, p = 0.007), male (t = –1.27, DF = 9, p = 0.24), and male/female (t = –3.04, DF = 19, p = 0.007) cohorts (paired T-test). (C) Interval plot comparing female sCPP% scores across sessions (One-way ANOVA, OWA, Dunnett’s post-hoc test, F = 262.45, p < 0.0001). (D) Interval plot comparing male sCPP% scores across sessions (OWA, Dunnett’s post-hoc test, F = 111.02, p < 0.0001). (E) Interval plot comparing male/female (all) sCPP% scores across sessions (OWA, Dunnett’s post-hoc test, F = 295.25, p < 0.0001| versus PrC, C3: p = 0.02 and Tst: p = 0.03). *p < 0.05, **p < 0.01.

Assessment of sCPP% score during the conditioning and test sessions showed the progression of reward context learning and retrieval of the learned biased context. Progression of sCPP task performance was assessed across experimental sessions for females, males, and male/female cohorts. Among females, the sCPP% score increased empirically from C2 and peaked with the test session (Figure 3C, one-way ANOVA, F = 262.45, p < 0.0001). However, there was no significant difference for sCPP% scores when all sessions were compared in a repeated measure One-Way ANOVA (OWA), with Dunnett’s post-hoc test. Male mice cohorts attained a comparable proficiency level at C3 (Figure 3D, OWA, F = 111.019, p < 0.0001). Together, the sCPP% score for the male/female cohort increased empirically at C2 and significantly at C3 (p = 0.02) and Tst (p = 0.03) sessions (versus the PrC, Figure 3E, F = 295.3, p < 0.0001).

Magnitude sCPP

Here, sucrose rewards were presented in both contexts during the conditioning (learning) phase (Figure 4A). A biased discrimination condition was set such that the magnitude of the reward in the PrC non-preferred context (target) was 10 times (20% sucrose) the weight of the PrC preferred context (non-target, 2% sucrose). Female mice showed a significant increase in sCPP% when the test phase was compared with the PrC (female:Figure 4B, t = −5.08, DF = 9, p = 6.57e-4). Similarly, male mice showed robust sensitivity to mag sCPP as the sCPP% for the test was significantly higher than the PrC score (male: Figure 4B, t = −5.41, DF = 9, p = 4.30e-4). A cohort of male/female mice also showed significantly higher test sCPP% than the PrC score (all:Figure 4B, t = −7.55, DF = 19, p < 0.0001).

sCPP% score for magnitude sCPP conditioned with 2%:20% sucrose. (A) Schematic illustration of magnitude (mag) sCPP with biased rewards conditioning. (B) Composite graph comparing preconditioning and test sCPP% scores for female (t = –5.08, DF = 9, p = 6.57e-4), male (t = –5.41, DF = 9, p = 4.30e-4), and male/female (t = –7.55, DF = 19, p < 0.0001) cohorts. (C) Interval plot comparing female sCPP% scores across sessions (OWA, Dunnett’s post-hoc test, F = 1132.348, p < 0.0001| versus PrC, C1: p = 0.048, C2: p = 3.4e-4, C3: p = 0.041, and Tst: p < 0.0001). (D) Interval plot comparing male sCPP% scores across sessions (OWA, Dunnett’s post-hoc test, F = 285.37, p < 0.0001| versus PrC, C1: p = 0.003, C2: p = 0.01, C3: p = 0.005, and Tst: p = 0.002). (E) Interval plot comparing all mice (male/female cohort) sCPP% scores across sessions (OWA, Dunnett’s post-hoc test, F = 876.5, p < 0.0001| versus PrC, C1: p = 1.06e-4, C2: p < 0.0001, C3: p = 1.52e-4, and Tst: p < 0.0001). *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Comparison of female mag sCPP% score across PrC, C1-C3, and Tst sessions (OWA RM, Dunnett’s post hoc) showed a significant increase in updated place preference from the first conditioning session, C1 (p = 0.048). The increased mag sCPP% was sustained for subsequent conditioning sessions C2 (p = 3.4e-4) and C3 (p = 0.041), and the test session (p < 0.0001) (Figure 4C, F = 1132.4, p < 0.0001). Similarly, compared to the PrC phase, male mice showed significantly higher mag sCPP% scores during C1 (p = 0.003), C2 (p = 0.01), C3 (p = 0.005), and the test session (p = 0.002) (Figure 4D, F = 285.37, p < 0.0001). Together, a combined group of male and female mice showed a significant increase in mag sCPP% score for C1 (p = 1.06 e-4), C2 (p < 0.0001), C3 (p = 1.52 e-4), and the test (p < 0.0001) sessions, when compared with the PrC outcomes (Figure 4E, F = 876.5, p < 0.0001).

Learning outcomes for reward magnitude discrimination are stronger than reward detection

In sCPP tests, context learning can be assessed during the conditioning phase. Across three biased conditioning sessions, the progression of contextual preference change can also be determined. Therefore, when the reward is unexpectedly omitted during the test sessions, memory retrieval related to biased conditioning can be further assessed. An increase in sCPP% score during the conditioning and test sessions indicates that the animal explored the target context more than they explored it in the PrC (i.e., sCPP%), as described in Figures 3, 4. For qualitative analysis of task performance, we examined the percentage of mice that recorded a target (Tg) time that is more than the non-target (nTg) exploration time during the test session. This qualitative analysis showed that the female cohort performed empirically better than males for abs sCPP and mag sCPP tasks. Likewise, both male and female cohorts have stronger empirical performance for mag sCPP (2%/20%), in comparison with abs sCPP (0%/16%) tasks. Thus, to assess changed decision during test sessions, the exploration duration of the PrC non-preferred side (Tg) must be greater than the duration in the initially preferred PrC context (nTg). In abs sCPP (Figure 5A), female mice recorded 40% of total changed preference during the test phase, while male mice recorded 20% changed preference. For the combined male/female cohort, there was a 30% (mean) change in preference during the test phase of the abs sCPP task. In comparison, female mice recorded an 80% change in preference during the test phase of mag sCPP tasks, while males had a 40% change in preference (Figure 5B). Together, the male/female cohort showed a 60% (mean) change in preference for the mag sCPP task.

Comparison of absolute and magnitude CPP factors (A,B). Pie charts showing the percentage distribution of mouse sessions with changed preference for absolute (abs) and magnitude (mag) sCPP tasks. (C) Schematic illustration of sCPP sensitivity based on modality (abs or mag). (D) Interval plot comparing abs and mag sCPP% scores for female mice. Significance (*) represents the difference between abs and mag sCPP% scores for the corresponding session (paired T-test, C1: t = –2.53, DF = 9, p = 0.03 and Tst: t = –2.42, DF = 9, p = 0.038). (E) Interval plot comparing abs and mag sCPP% scores for male mice (paired T-test, C1: t = –2.493, DF = 9, p = 0.03). (F) Interval plot comparing abs and mag sCPP% scores for the male/female (all) cohort. (paired T-test, C1: t = –3.65, DF = 19, p = 0.002, C2: t = –2.47, DF = 19, p = 0.02, and Tst: t = –2.78, DF = 19, p = 0.012). *p < 0.05, **p < 0.01.

These distribution patterns suggest that learning outcomes for mag sCPP are likely robust compared to abs sCPP in both male and female mice (Figure 5C), and no statistical difference was observed between male and female cohorts. The suspected difference in sensitivity to mag sCPP was further verified statistically by comparing the sCPP scores for the abs and mag experimental sessions. Among female mice, there was no significant difference in preconditioning sCPP% for abs and mag sCPP tasks. This result indicates that sensory cue changes across experiments did not alter preference exploration and performance outcomes (t = −0.17, DF = 9, p = 0.87, Figure 5D). Interestingly, mag sCPP scores during the first conditioning (C1, t = −2.53, DF = 9, p = 0.03) and test session (Tst, t = −2.42, DF = 9, p = 0.038) were significantly higher than abs sCPP scores. There was no significant difference between abs sCPP and mag sCPP scores during C2 (t = −2.103, DF = 9, p = 0.07) or C3 (t = −0.57, DF = 9, p = 0.58) (Figure 5D). Male preconditioning abs sCPP and mag sCPP scores were not significantly different (Figure 5E, t = 1.35, DF = 9, p = 0.21). For the male cohort, the mag sCPP score was significantly higher than abs sCPP during C1 (t = −2.49, DF = 9, p = 0.03), but not C2 (t = −1.49, DF = 9, p = 0.17), C3 (t = −0.96, DF = 9, p = 0.36), or the test (t = −1.53, DF = 9, p = 0.16) sessions. Combining males and females showed that mag sCPP is more sensitive than abs sCPP. In the learning (conditioning) phase, conditioning with 2%/20% sucrose produced stronger context preference changes when compared with the 0%/16% conditioning (Figure 5F). It is important to note that there was no significant difference in the PrC scores for mag sCPP and abs sCPP (t = 0.787, DF = 19, p = 0.44). Interestingly, mag sCPP outcomes were significantly higher than abs sCPP scores for C1 (t = 3.65, DF = 19, p = 0.002) and C2 (t = −2.47, DF = 19, p = 0.02) but not C3 (t = −1.04, DF = 19, p = 0.3). Ultimately, the mag sCPP test session score was significantly higher than the abs sCPP outcome (t = −2.78, DF = 19, p = 0.012).

Mechanism for dCA1 encoding of context

The behavioral tasks described above showed that mice have a higher sensitivity to reward magnitude discrimination (mag sCPP) compared to reward detection (abs sCPP) (Figures 5D–F). Behavioral sensitivity was derived as sCPP% and represents the threshold of preference change when the task is biased by a reward or a reward of higher magnitude. Given that the hippocampus contains a mix of cells that encode spatial orientation, reward probabilities, and other sensory cues, we propose that dCA1 neurons encode contextual preference at the population level. Within this framework, we determined whether the spatiotemporal coding pattern of dCA1 single unit population that represent an initial contextual preference is updated relative to a physical (behavioral) change in contextual preference. To test this proposition, we determined whether (i) a change in the firing rate, (ii) time of firing relative to context entry, and (iii) peak firing positions predict an initial contextual preference, and subsequently updated preference in sCPP tasks. Lastly, we determined if, and how, dCA1 single units respond to context discrimination paradigms that are driven by absolute and magnitude factors relative to the behavioral task outcomes described above. Spatiotemporal analysis of single units was performed to determine the relative location (i.e., context) where peak firing is attained. Within the hippocampus, reward/context-sensing cells were distinguished from place cells by their lower peak firing rate and change in peak firing position across sCPP task sessions.

Absolute sCPP task outcomes during dCA1 single unit recording

Mice with neural implants performed abs sCPP with four conditioning sessions (Figures 6A,B). After the preconditioning phase (without reward), three and four mice (n = 7), respectively, preferred context A and B (Figure 6C). To estimate task performance outcomes, preconditioning outcomes were reorganized such that all context preference (Pr) exploration duration was compared with all non-preference (nPr). The time spent exploring the preferred context was significantly higher than the non-preferred context exploration time (Figure 6D, t = −3.013, DF = 6, p = 0.024). This serves as the baseline abs sCPP% score, and the non-preference (nPr) was set as the target (Tg) for the subsequent sessions (C1–C4 and Tst). Across conditioning (C1–C4) and Tst sessions, the exploration time was comparable for target (Tg) and non-target (nTg) contexts (no significance). Mice with neural implants (Figure 2B) showed an empirical increase in abs sCPP% score when 16% and 0% sucrose were presented at the Tg and nTg contexts, respectively (Figure 6E, F = 83.68, p = 2.62e-4). Compared with the pre-conditioning (baseline), there was no significant change in sCPP% during C1-C3, and a significant increase at C4 (Figure 6E, OWA Dunnett’s post hoc, p = 0.047). Like the observation for mice (male) without implants, there was an empirical increase in abs sCPP% score when the Tst was compared with the PrC session (Figure 6E, F = 83.68, p = 2.62e-4).

Absolute sCPP outcomes for mice with neural implants. (A) Schematic illustration of rewards placement (0%:16%) for mice with context A or B preconditioning preference. (B) Schematic illustration of a mouse with neural implants during absolute (abs) sCPP task. (C) Graph showing time distribution in contexts A and B during the preconditioning phase. n = 3 mice preferred context A while n = 4 mice preferred context B. (D) Graphs comparing exploration time in the preferred and non-preferred contexts during the preconditioning session (paired T-test, PrC: t = –3.01, DF = 6, p = 0.024), then target and non-ta