The Hypothesis Dump

A young science such as neuroscience suffers from adolescent traits as it chases answers through technological development, deep data analysis, and big-data consortia. It lacks strong epistemological foundations and causal methodologies that would increase construct validity, and enable responses to fundamental questions, rather than mere indications of the direction of inquiry. There is, in my opinion, a need to supplement technological and analytical development with philosophical grounding. That would narrow our scientific methodology to a more concise, precise and accurate attempt at answering fundamental questions in Neuroscience.

Cajal’s neuron doctrine is approximately 130 years old, Donald Hebb’s approximately 80 years old; nevertheless, neuroscience is conceptually immature. Over the last few decades, neuroscience has developed a very powerful toolkit: optogenetics, connectomics, high-density recording, yet fundamental questions, such as how memory is stored and what a representation is, remain conceptually unresolved. Our hypotheses remain constrained not by reliance on technology and data harvesting but by the lack of epistemological exploration and methodological care. The field resembles an adolescent developmental stage: full of energy, lacking self-awareness, and mesmerised by novel instruments. This is unfortunate but not permanent. We may still affect the speed at which this maturation occurs by seeking advice from philosophy.

Optimally, we would start with a theoretical vacuum, measure, observe and draw conclusions. Unfortunately, we are left with borrowed historical baggage, psychological constructs that were inherited and not truly validated. The concern is not new as Skinner argued in 1950 that theories appealing to unobservable constructs were explanatory fictions (Skinner, 1950). His remedy, abandoning theory, was in my opinion as extreme as the problem he diagnosed. The warning went largely unheeded though, as the field’s instruments grew more powerful. Nowadays, paper after paper presents descriptive data analysis in correlation with phenomena, associating brain activity with mental states, and is often mistaken for explanation when it is at best an indication (Poldrack, 2006). The field rewards methodological measurement rather than meaning, resembling engineering more than science. Resolving fundamental questions requires experiments grounded in causal methodologies. Causation mechanically executed without epistemological rigour is exemplified by my own field of memory research, where, for instance, calcium imaging and electrophysiology are recorded without manipulation while behavioural output is monitored, followed by mechanistic explanations. Even when causal methodologies are implemented, such as optogenetic manipulation of neural circuits, off-target effects can confound functional interpretations (Otchy et al., 2015), potentially rendering them correlational.

These methodologies may be limited to descriptive understanding rather than elucidating the mechanisms of emergent properties. When neuroscience analytics are applied to systems we already understand, such as a microprocessor, they seem to fall short of yielding explanatory power (Jonas & Kording, 2017). Furthermore, the correlational enterprise is quantitatively fragile: typical brain-wide association studies require thousands of participants for reproducibility (Marek et al., 2022). If the aim is basic scientific inquiries, then this will leave us with large datasets that only point towards the more accurate hypotheses rather than elucidating underlying mechanisms.

Within my field of memory research, causal manipulation is possible but under-used relative to descriptive-scale enterprises, big-data consortia and brain atlases, for instance. Working on causal investigations may feel like swimming against the descriptive current. The attractiveness of big data and deep data analysis has driven the field toward a foundation for neuroscientific inquiry that is correlational. Moreover, deep, multilayered data analysis seems to have a siren-like allure for researchers, as it overrules experimental design and epistemological care, a manifestation of the reductionist, tool-driven bias the field has been warned against (Krakauer et al., 2017). The focus on big data is understandable in the search for diagnostic targets and treatments of pathology. However, for basic research questions, these data are insufficient, as they merely point to a direction of inquiry rather than explaining mechanisms.

As discussed in a previous article (Beyond Descartes), people have long contemplated the brain–mind relationship. We are nonetheless in need of a philosophical re-anchoring in neuroscience to deepen our epistemological exploration, hypothesis formation, and methodological precision. Such a re-anchoring can strengthen our construct validity and hone our practices toward more precise aims. For example, publishing hypotheses and methodologies for peer-review at the pre-experimental stage. This allows the community to evaluate the epistemological basis and methodological accuracy of each study and helps to uphold experimental precision for the highest explanatory power. This could build on emerging formats such as Registered Reports, extending review from methodological soundness to the epistemological grounding of the hypothesis itself. Furthermore, inviting collaborations with philosophy departments also at a pre-experimental stage to hone skills and help refine study constructs.

Declaration of AI use: AI (Lumo) was used for literature search, grammar, and structure editing. All scientific arguments, examples, and conclusions were written by the author.

Declaration of Interests: The author is the founder and editor of the site.

© 2026 Samer Siwani, The Hypothesis Dump. All original content is under a Creative Commons Attribution-Non Commercial- 4.0 International License. Privacy Policy | Terms of Service

References

Jonas, E. and Kording, K.P. (2017) ‘Could a neuroscientist understand a microprocessor?’, PLOS Computational Biology, 13(1), e1005268. doi:10.1371/journal.pcbi.1005268.

Krakauer, J.W., Ghazanfar, A.A., Gomez-Marin, A., MacIver, M.A. and Poeppel, D. (2017) ‘Neuroscience needs behavior: correcting a reductionist bias’, Neuron, 93(3), pp. 480–490. doi:10.1016/j.neuron.2016.12.041.

Marek, S., Tervo-Clemmens, B., Calabro, F.J., et al. (2022) ‘Reproducible brain-wide association studies require thousands of individuals’, Nature, 603(7902), pp. 654–660. doi:10.1038/s41586-022-04492-9.

Otchy, T.M., Wolff, S.B.E., Rhee, J.Y., Pehlevan, C., Kawai, R., Kempf, A., Gobes, S.M.H. and Ölveczky, B.P. (2015) ‘Acute off-target effects of neural circuit manipulations’, Nature, 528(7582), pp. 358–363. doi:10.1038/nature16442.

Poldrack, R.A. (2006) ‘Can cognitive processes be inferred from neuroimaging data?’, Trends in Cognitive Sciences, 10(2), pp. 59–63. doi:10.1016/j.tics.2005.12.004.

Skinner, B.F. (1950) ‘Are theories of learning necessary?’, Psychological Review, 57(4), pp. 193–216. doi:10.1037/h0054367.