**4. Discussion**

In the present study, we explored the degree to which imagery visualization and word familiarity affected learning outcomes in a traditional incidental recall paradigm. In doing so, the current study compared contrasting theoretical predictions regarding the relationship between the strength of neuropsychological variables and learning outcomes. Specifically, we contrasted associative theories that claim that the relative strength of neuropsychological variables should be additive [8], non-monotonic theories that claim that the relative strength of neuropsychological variables should be non-monotonic in relation to learning outcomes [9,10], and RSN theories which predict that learning outcomes should improve when competition between DMN and TPN is reduced [25,34,71].

Our main results demonstrate that recall probability is at its highest when the strength of our neuropsychological variables (noun familiarity/imagery vividness) are in a congruence at the time of stimulus encoding. In other words, when the strength of noun familiarity and imagery vividness are both strong or both weak, recall of the stimulus is at its highest probability. Using our dualistic RSN neuropsychological model, we were able to infer that the strength of our neuropsychological variables was associated with DMN activation when variables were of low strength and associated with TPN activation when variables were of high strength. When considering the inferred pattern of RSN activation, our results show that a competing activation pattern between DMN and TPN at the time of stimulus encoding results in significant impairment of recall probability.

An additional secondary finding was that the mean reaction time to complete the mental imagery task was quickest when the strength of our neuropsychological variables was high. In other words, it was quicker to generate visual mental images from noun cues under conditions where TPN was dominant at the time of encoding. The vivid-is-fast correlation between high image latency speed and high strength of neuropsychological variables has been reported in previous research [48,65,72,73]. However, our findings sugges<sup>t</sup> that in addition to high strength variables, dominant-TPN at the time of encoding is also a key predictor of quick imagery generation latency. These results sugges<sup>t</sup> an intriguing, novel explanation for the vivid-is-fast phenomenon. Namely, that it is not only the strength of neuropsychological variables that predict imagery generation latency but rather the dominant RSN pattern utilized during a mental imagery task. However, given the incongruent results between RSN patterns among our dependent variables (imagery latency was only significantly faster in the TPN grouping, while recall probability was significantly greater in both the DMN and TPN groupings), it is unclear whether image latency is related to the same processes responsible for memory consolidation. This is a research question open for future investigations.

In terms of the findings concerning the recall probability variable, our results do not support associative theories, as high memory consolidation is predicted to be exclusively a result of the additive strength of the preceding memory representations. Therefore, it can be concluded that Paivio's associative dual coding theory does not accurately model the incidental recall e ffect when operationalizing the strength of both mental imagery vividness and noun cue familiarity.

While our results did demonstrate a non-monotonic relationship between the strength of memory representations and recall probability, the directionality of our non-monotonic relationship did not align with that of Norman's non-monotonic-plasticity-hypothesis (NMPH). In the NMPH model, low strength memory representations result in lower memory consolidation scores compared to high strength memory representations. In our results, we found no significant di fferences between the high and low memory representations strength in terms of recall probabilities, suggesting that the NMPH may not accurately model memory consolidation at levels of reduction greater than single neurons. It is important to note that while our results did not precisely match the type of non-monotonic relationship that characterizes a neuron's level of excitation and the change in its associated synaptic strength, our results still demonstrated a non-monotonic relationship. This could sugges<sup>t</sup> that there

is, at the very least, a partial carryover e ffect of synaptic neuroplastic principles at the level of large scale distributed neural representations, given that non-monotonic results are rare within memory paradigms. However, the most likely interpretation of our results, that accounts for the largest amount of variability in our data, appears to be related to the variability in RSN dominance at the time of stimulus encoding. Utilizing the current RSN brain imaging literature [36,37] we were able to map the strength of our neuropsychological variables (imagery vividness/stimulus familiarity) with their associated dominant RSNs (DMN or TPN). This permitted the comparison of recall probability scores with the inferred dominant RSN at the time of encoding. Our data clearly sugges<sup>t</sup> that when there is an antagonist relationship between the inferred DMN and TPN activation at the time of stimulus encoding, recall probability, and therefore, memory consolidation is significantly impaired.

Our overall findings help to confirm a hypothesis emerging from the mental health literature suggesting that a possible causal mechanism of memory deficits in neuropsychiatric disorders, such as depression, anxiety, attention-deficit-hyperactivity disorder, autistic spectrum disorder, and schizophrenia, are characterized by abnormalities in RSN activation [24–34]. For example, in schizophrenia, there is a significant increase in DMN activity during attention-demanding tasks, which normally recruits the TPN in healthy controls [74]. The increased DMN activation in schizophrenic patients results in a metabolic competition between the two anti-correlated RSNs, which is hypothesized to explain their stunted performance on such tasks [75]. The strongest evidence of the RSN memory interference hypothesis originates from research on ADHD. In a recent randomized, double-blinded, placebo-controlled clinical trial, researchers explored the RSNs of both ADHD patients and healthy controls while they performed attention-demanding tasks. Their results demonstrated that ADHD patients were unable to suppress the appropriate RSN both during the task and in-between trials. More importantly, ADHD patients were shown to have a significant amount of RSN activation variability during the task. This increased RSN variability, or increased competition between DMN and TPN, is hypothesized by the researchers to be responsible for the reduced scores on the behavioral task compared to controls [25]. Critically, however, these clinical findings provide general evidence that RSN homeostasis is important for e fficient memory consolidation, but it is still unclear how abnormal RSN activation arises from such a wide variety of neuropsychiatric disorders. Hopefully, further experiments using healthy controls could help to determine the neurophenomenological marker's responsible for these core memory deficits.

To our knowledge, this study is the first to investigate how RSN variability might a ffect memory consolidation in healthy controls by manipulating the strength of neuropsychological variables that influence memory formation. While the results tend to validate the RSN memory deficit hypothesis in the mental health literature, there are still methodological limitations that should be considered prior to generalizing the results of this study. An initial limitation is a result of our reliance on indirect measurements of RSNs. While our dualistic RSN neuropsychological model is inferred from reliable brain imaging research [36,37], it is possible that the interaction between imagery vividness strength and stimulus familiarity at the time of stimulus encoding could reveal RSN activity which di ffers from our predicted inferences. Further studies using direct RSN imaging classification paradigms will need to be used to more precisely measure changes in dominant RSN activation in the function of both behavioral performance and of the strength of the chosen modulatory variables. A potential modulatory variable that could be used in future studies to help model the role of RSNs in memory consolidation is acute cannabis use. It has recently been demonstrated that cannabis is one of few pharmacological substances that has the e ffect of acutely increasing DMN connectivity during an attentionally demanding task [76]. Therefore, cannabis, in conjunction with psychological variables, such as mental imagery vividness and familiarly strength, could be used to artificially create an environment where individuals experience high degrees of competition between both the TPN and the DMN. Such experimental conditions could o ffer insight as to how various neuropsychological variables influence RSNs and therefore, memory consolidation processes.
