= Solution
\b[A reduced state space is useful when it preserves the structure needed for the question being asked; low dimension and high explained <variance> do not by themselves establish a sufficient neural code.] Start with a clear observation model. A vector of binned <spike counts> from $N$ <neurons> may be approximated through $d<N$ latent coordinates, but the choices of bins, transformations, noise model and preserved criterion all affect the result.
For centered observations, <principal component analysis> chooses an orthonormal projection $U\in\mathbb R^{N\times d}$ to minimize expected squared reconstruction error,
$$
E\lVert X-UU^TX\rVert^2=\operatorname{tr}(\Sigma)-\operatorname{tr}(U^T\Sigma U),
$$
where $\Sigma=E[XX^T]$. The identity follows by expanding the square and using $U^TU=I$. Diagonalize $\Sigma=V\operatorname{diag}(\lambda_i)V^T$. The trace term is $\sum_i\lambda_i\alpha_i$, with $\alpha_i=\lVert U^Tv_i\rVert^2$ for the columns $v_i$ of $V$. Here $0\leq\alpha_i\leq1$ and $\sum_i\alpha_i=d$ because the columns of $U$ are orthonormal. Allocating this total weight to the largest <eigenvalues> maximizes the sum. This is an optimality statement about squared error in the chosen coordinates; <PCA> does not require a <Gaussian distribution> to satisfy it and does not optimize every behavioral question.
An exact counterexample makes the limitation decisive. Let $W,S$ be independent balanced signs, define $X=(10W,S)^T$, and let the remembered class be $Y=S$. Zero means and <independence> give
$$
\Sigma=\begin{pmatrix}100&0\\0&1\end{pmatrix}.
$$
The first component retains $100/101\approx99.01\%$ of total <variance>. Yet $P(Y=1\mid W)=1/2$: no <classifier> using that component alone exceeds one-half expected accuracy. The second component determines the class exactly. Positive rates $(20+10W,5+S)$ give the same example after centering. These are constructed rates, not experimental recordings.
\Image[/past-exam-of-the-episodic-and-semantic-memory-course-of-the-university-of-oxford/2022-pca-memory-information.png]
{title=High explained variance can discard all information about a memory label}
This example also exposes a scaling assumption. Covariance-based <PCA> emphasizes large-variance coordinates; standardization changes that objective and may increase the influence of noisy low-rate cells. Neither preprocessing choice is automatically correct. Behaviorally relevant information may lie in a small-amplitude direction, in precise spike timing lost by binning, or in <correlations> not retained by a rate-only summary. Dimensionality must therefore be evaluated against decoding, reconstruction and the scientific question separately.
<Byron M. Yu> and colleagues' 2009 <Gaussian-process factor analysis> combined temporal smoothing and <dimensionality reduction> in a probabilistic framework. Its demonstration used 61 simultaneously recorded macaque premotor and motor <neurons> during reaching. It shows a method for estimating single-trial trajectories, not direct evidence that episodic memories have the same dimensional structure.<Gaussian-process factor analysis for low-dimensional single-trial analysis of neural population activity by Byron M. Yu and others>[2009, observation model, temporal priors and motor-recording demonstration]{ref}
In a linear observation model $y_t=Cz_t+b+\varepsilon_t$, the noise and latent temporal assumptions matter. A Gaussian observation approximation can be imperfect for sparse counts; independent residual errors can misassign shared fluctuations; a smooth temporal prior can suppress abrupt transitions. Trial averaging can manufacture a smooth path while concealing different single-trial paths. A useful low-dimensional description must also generalize beyond the sampled cells, session and task: unrecorded populations may encode variables absent from the fitted coordinates.
Latent axes require care as well. In $y=Cz+b$, an orthogonal rotation $Q$ gives the identical reconstruction through $z'=Qz$ and $C'=CQ^T$, because $Q^TQ=I$. Reconstruction alone therefore cannot label one axis a unique psychological process. Model priors and constraints can restrict this ambiguity. Similarly, attractive geometry does not establish intrinsic dynamics: unobserved inputs, behavior or the analysis itself may shape the apparent trajectory. If a true dynamical state is claimed, the retained coordinates should predict relevant future activity under a specified model, not merely provide a convenient picture.
Validation should choose dimension and preprocessing within <training data>, then assess held-out trials or sessions and, where appropriate, held-out <neurons>. Behavioral labels must be kept out of unsupervised fitting decisions used to claim independent prediction. Comparing multiple dimensions and reporting uncertainty is more informative than selecting the most compelling two-dimensional plot. A reduced representation can support a useful decoder without being a complete causal explanation of the circuit.
\b[Low-dimensional state spaces are testable approximations whose value depends on retained information, model fit and generalization. The projection, scaling, temporal and sampling assumptions should remain explicit, especially when a high-variance view is being interpreted as the mechanism of <episodic memory>.]
Back to article page
