You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: slides/week6/lecture20.html
+9-8Lines changed: 9 additions & 8 deletions
Original file line number
Diff line number
Diff line change
@@ -721,10 +721,10 @@ <h1 id="what-language-actually-does">What language actually does</h1>
721
721
<p>Stephens et al (2010) found that successful communication <strong>replicated the speaker's brain patterns in the listener's brain</strong>, with the listener's responses temporally coupled to — and sometimes <em>anticipating</em> — the speaker's. Language doesn't just convey ideas: <em>it programs one brain to enter the same state as another</em>.</p>
<p><ahref="https://doi.org/10.1073/pnas.1008662107"><strong>Stephens, Silbert & Hasson (2010, <em>PNAS</em>)</strong></a> "Speaker–listener neural coupling underlies successful communication" — Recorded brain activity from speakers and listeners during natural storytelling.</p>
724
+
<p><ahref="https://doi.org/10.1073/pnas.1008662107"><strong>Stephens, Silbert, & Hasson (2010, <em>PNAS</em>)</strong></a> "Speaker–listener neural coupling underlies successful communication" — Recorded brain activity from speakers and listeners during natural storytelling.</p>
<p>When person A watches a movie and then <em>tells</em> person B the story, person B's brain activity during listening resembles person A's brain activity during <em>watching</em> — not during speaking. Language doesn't just transmit verbal patterns. It reconstructs the speaker's <strong>perceptual experience</strong> in the listener's brain. You don't just hear a story — your brain <em>relives</em> it.</p>
<p>The brain constructs the same <strong>amodal narrative representation</strong> whether the story is read, heard, or watched as a movie. Language isn't special because of sound or letters — it programs a specific neural state regardless of input channel.</p>
<p><ahref="https://doi.org/10.1093/cercor/bhw351"><strong>Zadbood et al. (2017, <em>Cerebral Cortex</em>)</strong></a>: "How we transmit memories to other brains" — Narrative reconstructs the speaker's perceptual experience in the listener.</p>
737
-
<p><ahref="https://doi.org/10.1523/JNEUROSCI.1580-13.2013"><strong>Regev, Honey & Hasson (2013, <em>J. Neurosci.</em>)</strong></a>: Neural patterns are modality-independent.</p>
736
+
<p><ahref="https://academic.oup.com/cercor/article/27/10/4988/4080827"><strong>Zadbood et al. (2017, <em>Cerebral Cortex</em>)</strong></a>: "How we transmit memories to other brains" — Narrative reconstructs the speaker's perceptual experience in the listener.</p>
737
+
<p><ahref="https://doi.org/10.1523/JNEUROSCI.1580-13.2013"><strong>Regev, Honey, & Hasson (2013, <em>J. Neurosci.</em>)</strong></a>: "Selective and invariant neural responses to spoken and written narratives" — The same brain regions respond to stories regardless of modality.</p>
738
738
</div>
739
739
<divclass="tip-box" data-title="Discussion">
740
740
<p>If language reconstructs the speaker's perceptual experience in the listener's brain, is there a meaningful difference between "experiencing something" and "hearing a vivid enough description of it"? Where does the line fall? (Note: "hearing" could also mean "reading" or "watching" — the specific input channel isn't the point here.)</p>
@@ -749,8 +749,8 @@ <h1 id="llms-speak-the-brains-language">LLMs speak the brain's language</h1>
<p><ahref="https://doi.org/10.1016/j.neuron.2024.06.025"><strong>Zada et al. (2024, <em>Neuron</em>)</strong></a>: "A shared model-based linguistic space for transmitting our thoughts from brain to brain" — Demonstrated that LLM embeddings can track real-time speaker–listener neural coupling during natural conversation.</p>
751
751
<p><ahref="https://rdcu.be/cpMwZ"><strong>Heusser et al. (2021, <em>Nature Human Behaviour</em>)</strong></a>: "Geometric models reveal behavioral and neural signatures of transforming naturalistic experiences into episodic memories" — Brain activity during movie watching predicts how you <em>recount</em> it later.</p>
752
-
<p><ahref="https://doi.org/10.31234/osf.io/dh3q2_v2"><strong>Fitzpatrick et al. (2026, <em>Nature Commumnications</em>)</strong></a>: "Text embedding models yield high-resolution insights into conceptual knowledge from short multiple-choice quizzes" - Text embeddings can be used to accurately model (an approximation of) <em>everything</em> you know!</p>
753
-
<p><ahref="http://caligari.dartmouth.edu/~jmanning/pubs/MannKaha12.pdf"><strong>Manning & Kahana (2012, <em>Memory</em>)</strong></a>: "Interpreting semantic clustering effects in free recall" - When we use the "wrong" embedding model to model our thoughts and memories, how misleading is it?</p>
752
+
<p><ahref="https://doi.org/10.31234/osf.io/dh3q2_v2"><strong>Fitzpatrick et al. (2026, <em>Nature Communications</em>)</strong></a>: "Text embedding models yield high-resolution insights into conceptual knowledge from short multiple-choice quizzes" — Text embeddings can be used to accurately model (an approximation of) <em>everything</em> you know!</p>
753
+
<p><ahref="http://caligari.dartmouth.edu/~jmanning/pubs/MannKaha12.pdf"><strong>Manning & Kahana (2012, <em>Memory</em>)</strong></a>: "Interpreting semantic clustering effects in free recall" — When we use the "wrong" embedding model to model our thoughts and memories, how misleading is it?</p>
754
754
</div>
755
755
<divclass="tip-box" data-title="Discussion">
756
756
<p>Think about all <em>possible</em> concepts that "text" could express. LLMs cannot possibly learn this infinite space. Rather, they learn a <strong>much</strong> lower-dimensional subspace that captures something akin to "the concepts that are expressed in most text <em>in practice</em>." What are the implications of this idea? Does it speak to limitations of humans? Of language? Of text as a medium for thought and/or communication?</p>
@@ -820,7 +820,7 @@ <h1 id="what-llms-have-inside">What LLMs have inside</h1>
820
820
<p>Anthropic shows LLMs build structured concept maps. LeCun argues they lack grounding. Is a map of other people's concepts a form of understanding? Twist: as the Internet becomes "polluted" with LLM-generated text, how will this affect what LLMs can learn?</p>
<h1id="are-llms-thinking-or-performing">Are LLMs "thinking" or "performing"?</h1>
825
825
<divclass="definition-box" data-title="LLMs as role-play engines">
826
826
<p>LLMs are best understood as <strong>"engines for role-play"</strong> — they simulate the behavior of a plausible speaker, drawing on the vast repertoire of speakers in training data. When GPT writes a poem, it's not expressing itself — it's simulating someone who would write that poem. The performance can be indistinguishable from the real thing.</p>
@@ -829,7 +829,8 @@ <h1 id="are-llms-thinking-or-performing">Are LLMs "thinking" or "
829
829
<p>LLMs have <strong>attention</strong> (the mechanism) but may lack an <strong>attention schema</strong> — an internal model of their own attentional states. Humans don't just attend to things; we are <em>aware that we are attending</em>. LLMs process information without modeling the fact that they're processing it. They are "looking" but not "aware of looking."</p>
830
830
</div>
831
831
<divclass="note-box" data-title="References">
832
-
<p><ahref="https://doi.org/10.1145/3624724"><strong>Shanahan (2024, <em>CACM</em>)</strong></a>: "Talking about large language models" — LLMs as role-play engines simulating plausible speakers. <ahref="https://arxiv.org/abs/2411.00983"><strong>Farrell, Graziano et al. (2025, <em>arXiv</em>)</strong></a>: "Attention schema in LLMs" — LLMs lack an internal model of their own attentional states.</p>
832
+
<p><ahref="https://doi.org/10.1145/3624724"><strong>Shanahan (2024, <em>CACM</em>)</strong></a>: "Talking about large language models" — LLMs as role-play engines simulating plausible speakers.</p>
833
+
<p><ahref="https://arxiv.org/abs/2411.00983"><strong>Farrell, Graziano et al. (2025, <em>arXiv</em>)</strong></a>: "Attention schema in LLMs" — LLMs lack an internal model of their own attentional states.</p>
833
834
</div>
834
835
<divclass="tip-box" data-title="Discussion">
835
836
<p>When you read a novel and feel sad for a character, are you "really" feeling sadness, or performing a simulation of sadness triggered by text? If it's genuine for you, what would make it not genuine for an LLM?</p>
Copy file name to clipboardExpand all lines: slides/week6/lecture20.md
+10-8Lines changed: 10 additions & 8 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -55,12 +55,12 @@ Stephens et al (2010) found that successful communication **replicated the speak
55
55
56
56
<divclass="note-box"data-title="Further reading">
57
57
58
-
[**Stephens, Silbert & Hasson (2010, *PNAS*)**](https://doi.org/10.1073/pnas.1008662107) "Speaker–listener neural coupling underlies successful communication" — Recorded brain activity from speakers and listeners during natural storytelling.
58
+
[**Stephens, Silbert, & Hasson (2010, *PNAS*)**](https://doi.org/10.1073/pnas.1008662107) "Speaker–listener neural coupling underlies successful communication" — Recorded brain activity from speakers and listeners during natural storytelling.
59
59
60
60
</div>
61
61
62
62
---
63
-
<!-- _class: scale-70-->
63
+
<!-- _class: scale-65-->
64
64
65
65
# Language transfers memories between brains
66
66
@@ -78,9 +78,9 @@ The brain constructs the same **amodal narrative representation** whether the st
78
78
79
79
<divclass="note-box"data-title="Further reading">
80
80
81
-
[**Zadbood et al. (2017, *Cerebral Cortex*)**](https://doi.org/10.1093/cercor/bhw351): "How we transmit memories to other brains" — Narrative reconstructs the speaker's perceptual experience in the listener.
81
+
[**Zadbood et al. (2017, *Cerebral Cortex*)**](https://academic.oup.com/cercor/article/27/10/4988/4080827): "How we transmit memories to other brains" — Narrative reconstructs the speaker's perceptual experience in the listener.
[**Regev, Honey, & Hasson (2013, *J. Neurosci.*)**](https://doi.org/10.1523/JNEUROSCI.1580-13.2013): "Selective and invariant neural responses to spoken and written narratives" — The same brain regions respond to stories regardless of modality.
84
84
85
85
</div>
86
86
@@ -109,9 +109,9 @@ Text embedding spaces provide a **shared numerical coordinate system** for track
109
109
110
110
[**Heusser et al. (2021, *Nature Human Behaviour*)**](https://rdcu.be/cpMwZ): "Geometric models reveal behavioral and neural signatures of transforming naturalistic experiences into episodic memories" — Brain activity during movie watching predicts how you *recount* it later.
111
111
112
-
[**Fitzpatrick et al. (2026, *Nature Commumnications*)**](https://doi.org/10.31234/osf.io/dh3q2_v2): "Text embedding models yield high-resolution insights into conceptual knowledge from short multiple-choice quizzes" - Text embeddings can be used to accurately model (an approximation of) *everything* you know!
112
+
[**Fitzpatrick et al. (2026, *Nature Communications*)**](https://doi.org/10.31234/osf.io/dh3q2_v2): "Text embedding models yield high-resolution insights into conceptual knowledge from short multiple-choice quizzes" — Text embeddings can be used to accurately model (an approximation of) *everything* you know!
113
113
114
-
[**Manning & Kahana (2012, *Memory*)**](http://caligari.dartmouth.edu/~jmanning/pubs/MannKaha12.pdf): "Interpreting semantic clustering effects in free recall" - When we use the "wrong" embedding model to model our thoughts and memories, how misleading is it?
114
+
[**Manning & Kahana (2012, *Memory*)**](http://caligari.dartmouth.edu/~jmanning/pubs/MannKaha12.pdf): "Interpreting semantic clustering effects in free recall" — When we use the "wrong" embedding model to model our thoughts and memories, how misleading is it?
@@ -266,7 +266,9 @@ LLMs have **attention** (the mechanism) but may lack an **attention schema** —
266
266
267
267
<divclass="note-box"data-title="References">
268
268
269
-
[**Shanahan (2024, *CACM*)**](https://doi.org/10.1145/3624724): "Talking about large language models" — LLMs as role-play engines simulating plausible speakers. [**Farrell, Graziano et al. (2025, *arXiv*)**](https://arxiv.org/abs/2411.00983): "Attention schema in LLMs" — LLMs lack an internal model of their own attentional states.
269
+
[**Shanahan (2024, *CACM*)**](https://doi.org/10.1145/3624724): "Talking about large language models" — LLMs as role-play engines simulating plausible speakers.
270
+
271
+
[**Farrell, Graziano et al. (2025, *arXiv*)**](https://arxiv.org/abs/2411.00983): "Attention schema in LLMs" — LLMs lack an internal model of their own attentional states.
0 commit comments