Association profile conditioning in a set-temporal transformer for cross-session intracortical motor decoding

Xinyuan Zhang; Handong Mo; Pengfei Wen; Shuang Liang; Jichang Yang; Yan Zeng; Zhongrui Wang; Han Wang

q-bio.NCSep 30, 2026

Association profile conditioning in a set-temporal transformer for cross-session intracortical motor decoding

Authors: Xinyuan Zhang, Handong Mo, Pengfei Wen, Shuang Liang, Jichang Yang, Yan Zeng, Zhongrui Wang, Han Wang

Organizations: The University of Hong Kong, Hong Kong SAR, China · Southern University of Science and Technology, Shenzhen, China

Abstract

Intracortical motor decoders degrade across sessions because the set of recorded units changes and persisting units can alter how their firing relates to behavior. Most existing methods update network weights on each new session or rely on unlabeled activity, which does not directly reveal such changes. We present APST, an Association Profile-conditioned Set-Temporal transformer that adapts to new sessions with all network weights frozen. From a few labeled calibration trials, APST summarizes how each unit's firing relates to behavior in a four-dimensional association profile computed in closed form. The profiles condition a set-attention encoder that accepts any number and order of units, followed by a causal transformer for streaming decoding. On held-out DANDI688 sessions from two monkeys, APST reaches velocity $R^2$ of $0.78$ and $0.81$ , versus $0.40$ and $0.58$ for a variant that uses neural activity alone, and matches or exceeds an RNN fine-tuned on the same trials. On FALCON private held-out evaluation, it attains $R^2$ of $0.65$ , $0.42$ , and $0.44$ on M1, M2, and H1.

Figures & tables

Figure 1: Across sessions, electrode drift changes the recorded units, and a decoder fit on Day 0 degrades on Day $N$ without target calibration. APST adapts with frozen weights by conditioning on association profiles estimated from a few labeled calibration trials.

Figure 2: APST overview. (a) Target-session calibration: from

M

labeled trials, the association estimator produces each unit’s profile

p_{u}

; the profile conditions the activity signature at site \scriptsize1⃝ through FiLM and concatenation, forming unit identity

e_{u}

. (b) Streaming decoding: live activity is combined with the cached identity and profile at site \scriptsize2⃝, aggregated across units by set attention with learned slot queries, then decoded by a sliding-window causal transformer.

Method	Target adaptation	M1 ( $R^{2}$ / lat.)		M2 ( $R^{2}$ / lat.)		H1 ( $R^{2}$ / lat.)
Gradient-free target-session adaptation
WF	None	$0.34\pm 0.06$	0.06	$0.06\pm 0.04$	0.08	$0.16\pm 0.03$	0.15
RNN	None	$-0.60\pm 0.45$	0.03	$-0.07\pm 0.23$	0.01	$0.09\pm 0.18$	0.02
SPINT	Unlabeled few-shot	$0.66\pm 0.07$	0.13	$0.26\pm 0.13$	0.13	$0.29\pm 0.15$	0.14
APST (ours)	Labeled few-shot	$\mathbf{0.65\pm 0.11}$	0.15	$\mathbf{0.42\pm 0.10}$	0.09	$\mathbf{0.44\pm 0.15}$	0.15
Gradient-based target adaptation

Table 1: Official FALCON private held-out performance. Entries are mean

\pm

R^{2}

and normalized latency (lat.).

Figure 3: Results on DANDI688 and FALCON (shared y-axis: velocity

R^{2}

). (a) DANDI688 held-out sessions at 32 calibration trials. (b) Calibration curves (solid: Sub-C, dashed: Sub-M; blue: APST, pink: RNN-FT). (c) Ablations on DANDI688 held-out and FALCON dev. (d) Conditioning-site ablation on FALCON dev. (c,d) Three-seed means; negative bars are truncated with values annotated.

Method	M1	M2	H1
WF	$0.46$ ( $0.12$ )	$0.15$ ( $0.09$ )	$0.20$ ( $0.04$ )
RNN	$0.52$ ( $1.12$ )	$0.20$ ( $0.27$ )	$0.31$ ( $0.22$ )
SPINT	$0.77$ ( $0.11$ )	$0.59$ ( $0.33$ )	$0.47$ ( $0.18$ )
CycleGAN + WF	$0.61$ ( $0.18$ )	$0.32$ ( $0.10$ )	$0.15$ ( $0.03$ )
NoMAD + WF	$0.64$ ( $0.15$ )	$0.35$ ( $0.15$ )	$0.21$ ( $0.08$ )
NDT2 Multi	$0.77$ ( $0.18$ )	$0.63$ ( $0.20$ )	$0.62$ ( $0.10$ )

Table 2: Official private held-in mean

R^{2}

. Parentheses denote held-in minus held-out gap (

\mathrm{HI}-\mathrm{HO}

). Bold indicates our method.

Sub-C	11-13	11-16	11-17	11-19	11-20	12-01
APST	0.80	0.78	0.82	0.87	0.83	0.58
Act-only	0.56	0.58	0.65	$-0.02$	0.46	0.06
RNN-FT	0.83	0.81	0.83	0.83	0.85	0.48
Sub-M	06-23	06-25	06-26
APST	0.84	0.79	0.80
Act-only	0.49	0.60	0.66

Table 3: Held-out velocity

R^{2}

at 32 calibration trials. Act-only: Activity-only (no target labels); Dates in 2015 (same sessions as Fig. 3 , different seeds).

Explore similar work

Stable Neural Decoding Across Sessions via Task-Conditioned Latent Alignment for Brain-Machine Interfaces
Sep 23, 2026Canyang Zhao, Bolin Peng, J. Patrick Mayo +2Brain-Computer Interface Latent Representation Alignment
Institute of Automation, Chinese Academy of Sciences, Beijing, China · Departments of Ophthalmology and Bioengineering, University of Pittsburgh, Pittsburgh, USA · School of Science and Engineering, Chinese University of Hong Kong, Shenzhen, China
Test-Time Adaptation of Spiking Neural Networks for Intracortical Neural Decoding using Membrane Potential Alignment
Jun 12, 2026Guangzhi TangSpiking Neural Networks Brain-Computer Interface
Department of Advanced Computing Sciences, Maastricht University Maastricht, The Netherlands
End-to-End Intracortical Speech Decoding from Neural Activity
May 23, 2026Owais Mujtaba Khanday, Jose A. Gonzalez-Lopez, Marc Ouellet +2Character Error Rate Language Modeling
1Dpt. of Signal Theory, Telematics and Communications, University of Granada, Spain · 2Brain, Mind, and Behavior Research Center, University of Granada, Spain · 3Grupo CSUR de Epilepsia Refractaria, Hospital Virgen de Las Nieves (Granada), Spain

Sep 23, 2026, cs.LGJ/K move · Enter open · S save

Stable Neural Decoding Across Sessions via Task-Conditioned Latent Alignment for Brain-Machine Interfaces

Canyang Zhao, Bolin Peng, J. Patrick Mayo, Ce Ju+1

Institute of Automation, Chinese Academy of Sciences, Beijing, China · Departments of Ophthalmology and Bioengineering, University of Pittsburgh, Pittsburgh, USA · School of Science and Engineering, Chinese University of Hong Kong, Shenzhen, China

Brain-Computer Interface Latent Representation Alignment Controlled Comparison Long-Term Memory Sessions

Open paper AskPDF arXiv

Achieving stable long-term neural decoding in invasive brain-machine interfaces (BMIs) remains challenging due to variations in recorded neural populations across sessions. Current latent alignment approaches may overlook task-dependent structure during cross-session adaptation. We propose Task-Conditioned Latent Alignment (TCLA), a framework that stabilizes neural decoding by learning a shared latent space. TCLA learns a low-dimensional source representation using neural reconstruction and continuous behavioral supervision. During target-session adaptation, the shared representation is fixed, while target neural activity is mapped into the source latent space by aligning source and target distributions separately for each task condition. We evaluated TCLA on seven nonhuman primate datasets spanning multiple tasks. In long-term cross-session evaluation, TCLA achieved a mean

R^2

0.476\pm0.014

with a negative

R^2

failure rate of only 6.8%. Across 1,356 within-subject session pairs, TCLA achieved a mean

R^2

0.371\pm0.009

with a failure rate of 6.8%. Across 2,134 cross-subject session pairs, TCLA achieved a mean

R^2

0.218\pm0.004

with a failure rate of 12.9%, substantially better than those of the comparison methods. These results demonstrate that by preserving behaviorally relevant and task-dependent latent structure, TCLA improves the robustness of neural decoding across recording sessions and subjects. The source code is publicly available at https://github.com/FAMD-CASIA/TCLA.

Abstract

Explore similar work

Stable Neural Decoding Across Sessions via Task-Conditioned Latent Alignment for Brain-Machine Interfaces

Test-Time Adaptation of Spiking Neural Networks for Intracortical Neural Decoding using Membrane Potential Alignment

End-to-End Intracortical Speech Decoding from Neural Activity