The magnitude dimensions of visual stimuli, such as their numerosity, duration, and size, are intrinsically linked, leading to mutual interactions across them. However, it remains debated whether such interactions, or “magnitude integration” effects, arise from perceptual processes that are independent from the task performed, or whether they arise from high-level decision-making processes. We address this question with two electroencephalography (EEG) experiments in which participants watched a series of dot-array stimuli modulated in numerosity, duration, and item size, in two separate conditions. In the “magnitude task” condition, participants judged either the numerosity, duration, or size of each stimulus. In the “contrast task” condition, instead, a separate group of participants performed a contrast oddball task, never attending or judging the magnitude of the stimuli. The results of the magnitude task first show robust integration effects across the three dimensions. Then, we compare the neural responses to magnitude across the two task conditions. This comparison shows very similar brain responses irrespective of the task, within a series of latency windows whereby the modulation of response amplitude can predict the behavioral magnitude integration effect (~150 and ~250 ms post-onset for numerosity and size; ~300 ms post-offset for the effect of duration). To better assess the similarity of brain responses to magnitude irrespective of the task, we use a cross-condition multivariate decoding analysis. This analysis demonstrates that brain responses in the magnitude task can predict the responses in the contrast task, at multiple latencies starting from early processing stages (~120 ms). These results suggest that magnitude processing and integration likely involve perceptual processes that are engaged irrespective of the task, thus independently from decision making, although the effect of duration on other magnitudes may also involve post-perceptual processes such as working memory.