Prior to surgery, experimental animals were handled until calm and comfortable with the experimenter and habituated to the observer-side of the apparatus, where electrodes would eventually be tested. After surgery, but before any intracranial stimulation was given, all animals were habituated to the performer-side of the box on two occasions: once without the stimulating cable attached and a second session with the cable attached. After the efficacy of the electrode was tested in the observer-side of the box (see Electrode testing, below), no further habituation or exposure to the behavioral apparatus was given until the start of the experiments. Prior to the start of any subsequent session, regardless of the stage of training or testing, each animal was allowed to habituate to their compartment until they were calm, either sitting still or grooming. After each session, the box and manipulanda were cleaned thoroughly with detergent (Zalo Ultra, Lilleborg, Norway) and wiped dry to remove trace odors or cues that might influence subsequent experimental or training sessions.

After a minimum of 5 days of post-operative recovery, stimulating electrodes were tested for efficacy and the strength of stimulation required to reinforce behavior was determined for performers and observers. Efficacy was tested by reinforcing the animal’s preference for a neutral object (e.g. a pen) in the observation chamber of the experimental apparatus. During these tests, the animal was placed in the observation-side of the box and allowed to settle (1–5 min), after which single stimulations were delivered when the animal oriented toward or interacted with the object. All animals started at a stimulation intensity of 20µA. If they were non-responsive then current was increased incrementally by 2µA until behavioral effects were observed, such as increased investigation or physical interaction with the object. Current was then further increased in 2µA increments until side-effects were observed (such as motor artifacts or aversive reactions), or if there had been a cumulative increase of 10µA from the identified effective stimulation intensity without any observed side-effects. If movement artifacts were elicited by the simulation, current strength was incrementally lowered by 2µA until a stimulation intensity that did not elicit artifacts was identified. The final current strength was chosen from the upper range that elicited apparent reward without side-effects (ranging from 18 µA to 60µA across animals). If the optimal range was too narrow or unclear, the animal was re-tested and the optimum was determined on a subsequent day. Electrode testing for performers took place at the start of their training (described below in Training of performer rats), and for observers within 24 h before beginning the first observational session.

The overall procedure for training demonstrator rats consisted of three phases: (i) an initial shaping phase, followed by (ii) a continuous reinforcement schedule (i.e. stimulation delivered for every sphere-tap), which then changed to (iii) a partial reinforcement phase, in which reward was delivered only when the second sphere was tapped after the first sphere had been tapped without reward. At the start of the first shaping phase, MFB stimulation was given manually whenever the animals oriented toward or approached the spheres to encourage exploration. If no behavioral changes were observed after prolonged bouts of stimulation, or if the animals showed signs of aversion, stimulation current intensity was up- or down-adjusted, respectively. If the stimulation was still aversive or had no noticeable effect, training was discontinued and the rat was excluded from the study. After the animals showed sustained interest in either of the spheres, current intensity was again fine-tuned to the lowest current strength which yielded consistent behavioral responses.

Following initial shaping, subsequent training steps were structured as follows: (1) rewarding stimulation was given when the animal started to physically interact with either sphere, (2) they were rewarded only when starting to tap each sphere alternatingly (partial reinforcement of tapping behavior), (3) rewarded when tapping the spheres as in step 2, but only when the spheres were lit (with lighting controlled manually by the experimenter), (4) withholding reward when tapping the first sphere, but delivering reward when tapping the second sphere when cued by the light, (5) fully automatic training sessions until the animal performed > 75% successful trials per session. Step 4 and step 5 both relied on the Raspberry Pi controlled script to run the task and deliver instantaneous reward after the second tap, but differed in that step 4 allowed for extra motivational stimulation from the experimenter when the performer struggled with the task. Step 5 was initiated only after the animal toggled consistently back and forth between manually lit spheres without additional stimulation from the experimenter. Training was considered complete and stable once a performer exceeded 75% correct trials for three consecutive 30-minute sessions. Learning rates varied from animal to animal, and reaching criterion performance took anywhere from 2 to 15 training sessions. Naïve animals showed no initial spontaneous task-performance and typically required repeated stimulation to acquire the entire task-sequence. Performer rats used for subsequent experimental sessions were given a minimum of one day of rest after reaching criterion. The automatic training sessions utilized the same script as the subsequent experimental sessions.

During the experiments, observer and demonstrator rats were placed in their respective sides of the box and allowed to settle. Once the animals were calm the experimenter initiated the task and an automated script turned on the LED in the first sphere with a random time-interval between 3 and 30 s. The first light remained on for a maximum of 30 s if the demonstrator did not tap the sphere. If > 30 s elapsed, the LED turned off and the trial was scored as a “missed trial”, followed by a 3–30 s random time interval (inter-trial interval (ITI)) before the LED turned on again and a new trial started. If the animal tapped the first sphere within the 30 s when the light was on, the first LED turned off and the second LED in the other sphere turned on. The second LED also had a 30 s permissive time window. If the animal tapped the second sphere within this time window, the trial was scored as “successful”, and both the performer and observer received concurrent MFB stimulation as a reward. If the demonstrator animal did not tap the second sphere within 30 s, the trial was scored as “failed”. After each trial, whether successful or failed, another random ITI of 3–30 s was initiated before the next trial started and followed the same sequence as above. Each session lasted 30 min.

Observer animals observed either a well-trained performer or a naïve control performer (described below in Control group) for one 30-minute session per day over three consecutive days and were tested 24 h after the final observation session. The 24-hour delay was introduced to preclude spontaneous imitation (Zentall 2006). During observational sessions, observers were able to see the performer for the entirety of each 30-minute session while allowed to move freely in their side of the box. On the day of testing, observers were placed in the performer-side of the box, and the same pre-programmed LED- and MFB stimulation-script was run as when demonstrators performed the task. No pre-training or priming with stimulations was given to the observers before testing.

The control condition was run identically as the real experiments, but with performer animals that were completely naïve to the task. During each of the three control observation sessions, the naïve performer was allowed to move freely on the demonstrator side of the box while a custom script drove the spheres to light up and extinguish in the correct sequence, with reward stimulation delivered to both animals when the LED in the second sphere turned off. The lighting of the spheres and MFB stimulation progressed automatically, irrespective of the demonstrator’s actions, and trials were started at random and followed the same randomized 3 to 30 s ITIs as with the experimental trials. This way, reward delivery was dissociated from the behavior of the performer but maintained the sequential structure of correct trials. Critically, this condition also provided the social cue of a would-be performer but lacked the demonstration of task-specific behavior in conjunction with the reward.

After experiments were completed, animals were deeply anesthetized with Isofluorane and injected intraperitoneally with an overdose of pentobarbital (Exagon vet., 400 mg/ml, Richter Pharma Ag, Austria), after which they were perfused intracardially with 0.9% saline and 4% paraformaldehyde. The animals were then decapitated, and skin and muscle were removed from the skull before leaving it to post-fix overnight in 4% paraformaldehyde. The following day the electrode implants were removed from the skull and the brains were extracted and stored in dimethyl sulfoxide (DMSO) at 4 °C. On the day of sectioning, the brains were removed from DMSO, frozen with dry ice and sectioned with a sliding microtome at 40 μm in the coronal plane in three series. The first series was mounted immediately on glass slides, Nissl-stained and cover slipped, and the other two series were kept for long term storage in DMSO at -25 °C. Electrode placement was confirmed using the Nissl-stained series (Fig. 2). Of 23 total rats implanted for the final version of this paradigm, two were excluded due to electrodes being off the intended target, and 5 animals had correctly placed electrodes but were excluded due to disruptive behavior (e.g. excessive jumping) on the day of testing.

Histological verification of electrode placement. A Nissl-stained tissue section showing the termination point of a bipolar stimulating electrode targeting the medial forebrain-bundle. Scalebar = 2000 μm