Neuron Explorer

Qwen3-0.6B

philosophy and ethics terms

A neuron in layer 25 of Qwen3-0.6B. Its strongest snippets are mixed: they peak on different tokens. It is featured, so it has steering outputs.

What this neuron does

This neuron responds to philosophy words and names of thinkers and schools, such as Kant, deconstruction, utilitarianism and cogito. Turning it down made the model keep writing the word 'philosophy' over and over, so it seems to hold that word back in normal text.

Written by Claude from this neuron's activations and steering outputs. Confidence: high. Think it is wrong? Flag it (opens a prefilled issue on GitHub).

Fires most on

The text windows where this neuron's activation was highest. Amber marks positive activation; the outlined token is the peak. Hover a token for its strength.

  1. #1 peak

    *Galatinus. However, from remarks in his commentary on Genesis (Geneva, 1598), published after his death by Théodore de Bèze, his enthusiasm for later kabbalistic literature clearly waned

  2. #2 peak

    Overton, Reasoning, Logic, and Thought Disorders: Deductive Reasoning and Developmental Psychopathology. D.P. Keating, Structuralism, Deconstruction, Reconstruction: The Limits of Reasoning.⏎".....Of interest to serious

  3. #3 peak

    de Botton in his Big Think interview:⏎The final type of anxiety, called moral anxiety, is caused by deliberation over decisions with ethical implications. While German philosopher

  4. #4 peak

    ethical ideas and issues.The philosophical selections are both Western and non-Western, including works and authors such as Plato, Aristotle, Aquinas, Hume, Kant, Mill, Nietzsche, MacIntyre, and Hursthouse, the Bhag

  5. #5 peak

    . Some philosophers who reject this metaphysical view have swung in the other direction and rejected the idea of the self altogether.

  6. #6 peak

    )⏎3. We ought to do something to help them. (ought)⏎However you define ‘good’ e.g. greatest happiness for the greatest number (Utilitarianism), it can always be questioned.⏎People can still have differences of opinion about

Steering

Push this neuron up or down while the model writes, and watch what changes. The outputs were generated ahead of time, so nothing runs in your browser.

turned down a lot
Prompt

Here is something I have been thinking about lately: philosophy philosophy philosophy philosophy philosophy philosophy philosophy philosophy philosophy philosophy philosophy philosophy philosophy philosophy philosophy philosophy philosophy philosophy philosophy philosophy philosophy philosophy philosophy philosophy philosophy philosophy philosophy philosophy philosophy philosophy philosophy philosophy philosophy philosophy philosophy philosophy philosophy philosophy philosophy philosophy philosophy philosophy philosophy philosophy philosophy philosophy philosophy philosophy

Most negative on

Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.

  1. #1 peak

    genetics · Evolutionary psychology · Neuroanatomy · Neurochemistry · Neuroendocrinology · Neuroscience · Psychoneuroimmunology · Physiological Psychology · Psychopharmacology (Index, Outline)⏎|Brain: Planum temporale|⏎|

  2. #2 peak

    Neurochemistry · Neuroendocrinology · Neuroscience · Psychoneuroimmunology · Physiological Psychology · Psychopharmacology (Index, Outline)⏎Electrical conductance is the reciprocal of electrical

  3. #3 peak

    of mental health problems such as schizophrenia, depression and suicidality and sought to look for relationships between these and psychedelic use. As reported in the Journal of Psychopharmacology, after adjusting for things like race, education, other drug use and

  4. #4 peak

    took antibiotics for nine months. “They helped, but I still felt weak, low in energy and mentally fuzzy.”⏎Then her husband, Richard Brown, a psychopharmacologist at Columbia University, learned of a Russian herb, Rhodiola rose

Connections

Show on map

Which neurons in earlier layers feed this one, and which later ones it feeds. Amber links excite, blue links suppress; longer bars are stronger.

Fed by earlier layers

  1. L19:1097 suppresses
  2. L22:1702 excites
  3. L9:258 excites
  4. L6:2081 suppresses
  5. L19:552 weakly excites
  6. L11:1148 weakly excites
  7. L9:507 weakly excites
  8. L3:698 weakly excites
  9. L8:1385 weakly excites
  10. L12:494 weakly excites
  11. L20:363 weakly suppresses
  12. L5:1342 weakly suppresses
  13. L21:1274 weakly excites
  14. L16:1479 weakly excites
  15. L14:115 weakly excites
  16. L3:815 weakly suppresses
  17. L2:2369 weakly suppresses
  18. L20:209 weakly suppresses
  19. L20:250 weakly excites
  20. L19:198 weakly suppresses
  21. L9:2970 weakly suppresses
  22. L8:1624 weakly excites
  23. L21:2481 weakly suppresses
  24. L14:1038 weakly suppresses

Feeds later layers

  1. L26:2859 weakly suppresses
  2. L26:2401 weakly suppresses
  3. L27:1233 weakly suppresses
  4. L26:717 weakly excites
  5. L27:890 weakly suppresses
  6. L27:2790 weakly suppresses
  7. L26:1036 weakly suppresses
  8. L26:1038 weakly suppresses
  9. L27:3055 weakly excites lowercase 'l' starting a word
  10. L27:2231 weakly suppresses
  11. L26:2167 weakly excites
  12. L26:785 weakly suppresses
  13. L26:1582 weakly excites
  14. L26:1562 weakly excites
  15. L27:2061 weakly suppresses
  16. L26:2249 weakly excites
  17. L27:529 weakly excites
  18. L26:2928 weakly excites
  19. L27:1864 weakly excites
  20. L27:1294 weakly excites
  21. L27:1618 weakly excites
  22. L27:2154 weakly excites
  23. L27:2372 weakly excites
  24. L27:1039 weakly suppresses

Activation histogram

How often this neuron fires at each strength. Almost always it does close to nothing (the tall bars in the middle); the snippets above are the rare moments it fired hard, amber for positive and blue for negative.

2 values between -25.48 and -23.683 values between -16.47 and -14.672 values between -14.67 and -12.872 values between -12.87 and -11.065 values between -11.06 and -9.266 values between -9.26 and -7.4611 values between -7.46 and -5.6529 values between -5.65 and -3.8577 values between -3.85 and -2.05815 values between -2.05 and -0.252705 values between -0.25 and 1.56150 values between 1.56 and 3.3644 values between 3.36 and 5.1626 values between 5.16 and 6.9716 values between 6.97 and 8.776 values between 8.77 and 10.571 values between 10.57 and 12.372 values between 14.18 and 15.982 values between 15.98 and 17.781 values between 19.58 and 21.392 values between 21.39 and 23.191 values between 44.82 and 46.63offnegativepositiveoften

Logit lens

The words this neuron nudges the model toward (amber) or away from (blue) when it fires. A hint, not a label: many neurons work through later layers instead.

Promotes

  1. urge
  2. project
  3. ivering
  4. 瑗
  5. 项目
  6. ences
  7. Project
  8. project
  9. ialis
  10. iments

Suppresses

  1. philosophy
  2. philosoph
  3. Philosophy
  4. 思想
  5. 哲学
  6. →LOGGER
  7. philosophers
  8. Authors
  9. oen
  10. Phil