Selective activation of the M subtype of muscarinic acetylcholine receptor, via positive allosteric modulation (PAM), is an exciting strategy to improve cognition in schizophrenia and Alzheimer's disease patients. However, highly potent M ago-PAMs, such as MK-7622, PF-06764427, and PF-06827443, can engender excessive activation of M, leading to agonist actions in the prefrontal cortex (PFC) that impair cognitive function, induce behavioral convulsions, and result in other classic cholinergic adverse events (AEs). Here, we report a fundamentally new and highly selective M PAM, VU0486846. VU0486846 possesses only weak agonist activity in M-expressing cell lines with high receptor reserve and is devoid of agonist actions in the PFC, unlike previously reported ago-PAMs MK-7622, PF-06764427, and PF-06827443. Moreover, VU0486846 shows no interaction with antagonist binding at the orthosteric acetylcholine (ACh) site (e.g., neither bitopic nor displaying negative cooperativity with [H]-NMS binding at the orthosteric site), no seizure liability at high brain exposures, and no cholinergic AEs. However, as opposed to ago-PAMs, VU0486846 produces robust efficacy in the novel object recognition model of cognitive function. Importantly, we show for the first time that an M PAM can reverse the cognitive deficits induced by atypical antipsychotics, such as risperidone. These findings further strengthen the argument that compounds with modest in vitro M PAM activity (EC > 100 nM) and pure-PAM activity in native tissues display robust procognitive efficacy without AEs mediated by excessive activation of M. Overall, the combination of compound assessment with recombinant in vitro assays (mindful of receptor reserve), native tissue systems (PFC), and phenotypic screens (behavioral convulsions) is essential to fully understand and evaluate lead compounds and enhance success in clinical development.