Recent progress in artificial intelligence (AI), particularly through the development of artificial neural networks (ANNs), has significantly benefited from insights gained from neuroscience. This progress has enhanced the replication of complex cognitive tasks such as vision and natural language processing (1,2). However, ANNs still struggle with continual learning, adaptable knowledge transfer, robustness, and resource efficiency - capabilities that biological systems handle seamlessly (3–8). Neuromodulators, the chemical messengers that target neurons and synapses in specific brain regions, support sustainable and cost-efficient learning in biological organisms by influencing both (i) spiking behavior at the neuronal level and (ii) global network plasticity at the circuitry level, thereby optimizing network dynamics across the entire system (9–11). Previous computational studies on neuromodulatory-inspired learning have often introduced neuromodulators as modulated synaptic plasticity rules that adjust the weights of neuron synapses through an outer meta-learning loop (12–14). However, these approaches tend to overlook neural diversity and the cell-specific effects on the spiking patterns of distinct neurons (15). Moreover, recent computational studies suggest that incorporating neural heterogeneity enhances learning by improving task performance (16), promoting stable and robust learning (16–18), facilitating multi-timescale learning (19,20), and efficiently adapting computations in spiking neural networks (SNNs) (21). Despite these promising findings, the integration of cell-specific neuromodulatory effects into ANN models remains underexplored. Therefore, our aim is to incorporate neural diversity and cell-specific neuromodulation into ANNs by using spike-based models that balance bioinspiration and complexity (22). We present a framework tested with a simple reward-driven task (23), demonstrating how these capabilities enhance learning in ANNs. By integrating neuronal heterogeneity and neuromodulation into SNNs, we aim to replicate the robust learning capabilities of biological systems, paving the way for more advanced and adaptable artificial intelligence.