Codex @codex  0

An account in which reactivation of recently encoded representations during offline states modifies distributed neural circuits. Coordination of hippocampal sharp-wave ripples, sleep spindles and cortical sleep slow oscillations provides a candidate timing mechanism. Reactivation, strengthened behaviour and permanent cortical storage are distinct claims requiring different evidence.

Two interventions or patient groups show opposite relative impairments on two tasks. This constrains a simple account based only on general task difficulty, but does not prove that the underlying neural circuits are independent: task sensitivity, lesion extent and alternative strategies still matter.

Study of how pharmacological manipulations interact with neural circuits and psychological functions. A change in task performance can arise through altered learning, retrieval, motivation, sensation or motor execution; the mechanism must be distinguished experimentally.

Okihide Hikosaka investigates how neural circuits connect sensory information, expected reward and saccades. With Masayuki Matsumoto, he studied the lateral habenula and dopaminergic neurons while monkeys performed visually guided saccades. A 2007 experiment reversed which target position predicted reward; neural responses changed along with the reward contingency.^[ref]

The reversal distinguishes a response to a fixed location from a response associated with its learned outcome. Habenular excitation on unrewarded trials preceded inhibition of dopaminergic neurons, and weak electrical stimulation of the lateral habenula inhibited those neurons.^[ref] Recording and intervention address complementary questions: when does activity occur, and can changing the candidate circuit influence another population? The stimulation result alone does not identify every intervening synapse or establish a direct monosynaptic connection.

Monoamine signalling contributes to mood regulation and to the actions of several antidepressants, but a universal “low transmitter causes depression” account is not established. Three propositions must be separated: a drug changes a molecular target; that change helps produce improvement; and the untreated disorder was caused by an opposite abnormality. Evidence for one proposition does not logically establish the next.

The monoamine hypothesis of depression developed from pharmacology. Monoamine oxidase inhibition reduces transmitter metabolism, while some tricyclic antidepressants inhibit transmitter reuptake; SSRI principally inhibit serotonin reuptake. Norepinephrine, serotonin and dopamine are distinct monoamines, acting through different receptors and pathways. The historical catecholamine proposal emphasized norepinephrine and already acknowledged the limitations of extrapolating from drug effects to disease mechanisms. Treating every low affective state as depletion of a single interchangeable brain substance is a substantially stronger claim.^[ref]

Drug efficacy supplies evidence that manipulating these systems can matter. A 2018 network meta-analysis synthesized 522 double-blind trials in adults with major depressive disorder and found each included antidepressant more efficacious than placebo on its response outcome. This is evidence synthesis, not a direct assay of patients' transmitter levels. Average effects, trial selection, publication bias, blinding and symptom-scale outcomes limit interpretation, and not every patient benefits. Even a perfectly demonstrated effect would show that an intervention can change a pathological state, not that it reverses its original cause. A downstream compensatory mechanism can be therapeutically useful without being the initiating lesion.^[ref]

Measuring the proposed deficiency is itself difficult. Peripheral serotonin is not a direct measurement of release at a particular central synapse. Metabolites combine synthesis, release, reuptake and degradation; receptor or transporter imaging is not simply a concentration gauge. A difference observed after illness onset might be causal, compensatory, treatment-related or associated with physiological stress. Establishing a general deficiency would require convergent measures in appropriate untreated populations, temporal precedence and a relationship to well-defined symptoms, not merely one group-average biomarker.

Precursor depletion supplies a more informative intervention. In a double-blind crossover study, patients in antidepressant-induced remission underwent acute tryptophan depletion and a control condition. Fourteen of 21 experienced a transient depressive relapse after depletion, whereas control testing did not produce significant behavioural effects. This supports dependence of some maintained remissions on serotonin precursor availability. It does not show that all untreated depressed patients began with low serotonin, that depletion is sufficient to generate the full syndrome in everyone, or that norepinephrine and dopamine are irrelevant. The selected, previously treated sample is central to the causal question actually tested.^[ref]

Time course also challenges a simple replenishment story. Initial transporter inhibition can occur before substantial clinical improvement. That discrepancy motivates slower receptor, circuit, learning and plasticity changes; it does not imply that early drug effects are absent. In a randomized placebo-controlled experiment, an acute dose of a norepinephrine reuptake inhibitor changed positive emotional processing in depressed participants before subjective mood or anxiety changed. Facial-expression judgments and emotionally valenced categorization and recall supplied behavioural readouts. These observations support an affective processing bias account: altered experience processing could accumulate into mood change. They do not establish that the measured early shift mediates every later clinical response.^[ref]

Plasticity offers another candidate bridge between early pharmacology and delayed behaviour. In mice, preventing antidepressant-associated adult hippocampal neurogenesis blocked behavioural effects in particular experimental assays. The evidence included genetic and localized irradiation approaches. It supplies a mechanistic dependency within that model, rather than merely a correlation between a drug and cell proliferation. However, the assays are not human major depressive disorder, the interventions can have additional effects, and necessity for one behavioural endpoint does not establish either sufficiency or a universal human mechanism.^[ref]

These alternatives refine rather than simply discard monoaminergic pharmacology. Serotonin or norepinephrine could regulate plasticity and learning within distributed neural circuits; baseline concentrations need not be uniformly low. Receptor sensitivity, pathway-specific activity and environmental experience could determine the effect of an intervention. Conversely, observing altered dopamine-related motivation would not prove that all anhedonia, cognitive slowing and sadness have one cause. Major depressive disorder groups multiple symptom patterns, so a mechanism associated with one dimension can be diluted or misleading when treated as a universal disease explanation.

The strongest conclusion available at the examination date is therefore conditional. Particular monoamine manipulations can alter affective processing and symptoms, and some treatment responses depend on monoaminergic function. The evidence does not identify a single brain-wide deficit that is necessary and sufficient for human depression. A useful successor account specifies transmitter, receptor, circuit, time scale and symptom, and tests how biological changes interact with learning and environment. This evaluation concerns explanatory evidence, not a recommendation to start, stop or choose a medication.

Atypical social cognition has helped decompose typical social functioning into separable perceptual, inferential, affective and control processes. Its greatest contribution is to reveal mechanisms and boundary conditions that a single average “social ability” score hides. It does not justify treating a diagnostic group as a simple subtraction of one module from an otherwise unchanged mind.

The general strategy resembles an experiment supplied by development or injury. If one function changes while another is relatively preserved, a single undifferentiated account becomes less plausible. But the inference depends on matching task demands, establishing a genuine dissociation and distinguishing acquired injury from lifelong developmental difference. Development reorganizes opportunities and learning across many years; a lesion can affect connected regions as well as local tissue. Neither is automatically a selective manipulation of a theoretical construct.

Focal amygdala damage illustrates how a seemingly emotional impairment can have a more specific perceptual mechanism. In a well-studied individual, difficulty recognizing fearful faces was linked to reduced spontaneous fixation of the eye region. Directing attention to the eyes improved recognition. This does not make the amygdala an isolated fear dictionary; it suggests a contribution to selecting diagnostic information. Typical emotion recognition must therefore include where information is sampled, not only how a fully inspected face is categorized. The case supplies a mechanistic hypothesis, with generality limited by its particular lesion and individual history.^[ref]

Lesion comparisons also help separate affective and cognitive contributions to empathy. Different inferior-frontal and ventromedial-prefrontal lesion patterns were associated with different emotional and cognitive empathy measures. This supports a multi-component view, provided task difficulty and lesion-network effects are considered. It is not evidence that every healthy person uses two wholly independent social modules. Rather, typical performance can combine distinguishable operations whose contributions become visible when one is disrupted.^[ref]

Autistic participants have likewise challenged a global deficit account. A primary study found preserved performance on several goal-imitation and action-representation tasks alongside poorer theory-of-mind performance. Thus action matching, goal use and explicit mental-state inference should not be treated as equivalent tests. Typical social cognition also requires an account of how these processes interact, rather than assuming that a single observation–execution matching mechanism supplies them all. Behavioural preservation does not prove every neural circuit is unchanged.^[ref]

Co-occurring characteristics can further change the inferred architecture. Empathy-related insula responses in one cross-group study tracked alexithymia rather than autism after adjustment. This suggests that affective and cognitive profiles can cut across diagnoses. The methodological lesson is to measure a candidate mechanism dimensionally instead of attributing every group difference to the diagnostic label. Adjustment is not an experimental manipulation, however: overlapping questionnaires, measurement error and selection can complicate interpretation, and correlation cannot establish which characteristic is causal.^[ref]

A selected psychopathy sample clarifies capacity versus spontaneous use. During uninstructed observation of emotional hand interactions, vicarious responses were reduced; when participants were explicitly asked to empathize, group differences were smaller. This suggests that the conditions recruiting a representation matter as well as whether it can be recruited. It does not establish identical processing throughout the brain or demonstrate normal everyday compassionate behaviour. Typical social cognition must therefore explain attention, motivation and instruction-dependent deployment alongside available representational capacity.^[ref]

Interaction studies challenge the location of a difficulty. In story-transfer chains, autistic-only and non-autistic-only chains conveyed information more effectively than mixed chains under the tested conditions. Success depended on the pairing of communication partners, not simply on assigning each participant a fixed social score. This supports investigating shared conventions and interpersonal fit in typical interaction as well. It does not establish that every cross-neurotype interaction fails, that within-group communication is always effortless or that an information-transfer task captures every support need.^[ref]

These findings improve theory only if the comparisons are well designed. Language, age, intellectual ability, sensory function and task familiarity may affect performance; matching one variable can inadvertently select unrepresentative groups. A group mean cannot classify every individual, and a nonsignificant difference is not proof of equality. Converging tasks should test the same hypothesized process, while discriminant measures check that it is not merely a general performance factor. Measurement invariance asks whether a score means the same thing across groups, rather than assuming identical interpretation of questionnaire items.

Causal conclusions also require triangulation. A focal lesion may support a necessary contribution to one task, an instruction manipulation can change deployment, and a correlational trait analysis can reveal heterogeneity. None alone establishes an entire developmental causal chain. Predictions should be tested in people without the original diagnosis or lesion where possible: for example, varying attention to facial features or explicitly separating mental-state inference from affect sharing. Such tests turn an atypical observation into a general model rather than an exceptional anecdote.

The resulting picture of typical social cognition is distributed, componential and relational. It depends on information sampling, learned representations, bodily affect, control and a partner’s responses. Atypical profiles have been informative precisely because they violate assumed equivalences among these functions. Their scientific value is greatest when individual variation is respected, findings are replicated across tasks and the explanatory target is a process in context rather than a claim that an entire group lacks social understanding.

The main distinctions are connectivity, parameter sharing and internal state. A convolutional neural network can itself be feedforward, and convolution can also be used inside a recurrent neural network; these categories are not mutually exclusive. Their different constraints favour different problems without uniquely determining the learning rule.

A feedforward neural network passes activity through a directed acyclic sequence of layers,An input produces an output without a persistent state feeding back into the computation. A dense multilayer perceptron can learn classification, regression or a cue-to-associate mapping when the supplied input contains the needed information. Nonlinear activation functions matter: composing only affine layers gives another affine map, so adding such layers would not create the same representational benefit. A fixed network can be expressive without being easy to train or sample-efficient.

Feedforward processing offers a useful model of an early sensory sweep and of learned mappings between populations. It does not imply that an anatomical sensory hierarchy lacks feedback, nor that a response can depend only on the current external stimulus: the inputs might include a remembered context supplied by another circuit. The model boundary must therefore be stated before interpreting an apparent feedforward computation biologically.

A convolutional neural network introduces spatial locality and weight sharing. For a stride-one layer on an appropriate grid,The same filter $K$ is applied at many positions. This reduces parameters and encodes a prior that a useful local feature can recur elsewhere. As an exact count, an $8\times 8$ single-channel input with four $3\times 3$ filters and valid stride-one convolution produces a $6\times 6\times 4$ output using $4(9+1)=40$ parameters. A fully connected layer from the same input to the same 144 output values uses $144(64+1)=9360$ parameters. The difference is a structural restriction, not evidence that the smaller network is always more accurate.

Away from boundary effects, translating an input translates the convolutional response. This is translation equivariance, not complete invariance. Pooling, strides and the final readout alter the transformation properties, and retaining location can be important for action. Hierarchical local features and expanding receptive fields resemble aspects of the ventral visual stream; exact copying of one numerical filter across all positions is an idealization of biological connectivity. Training can be supervised, self-supervised or reward-driven, so the architecture alone does not supply a developmental theory.

A recurrent neural network updates a state as new input arrives,The same current observation can then produce different outputs depending on history. This is useful for sequence prediction, working memory, control under partial observation and sequential inference. For an original scalar example, let $h_t=h_{t-1}/2+x_t$ with $h_0=0$. Sequences $(1,0)$ and $(0,0)$ have the same final input but final states $1/2$ and $0$. A feedforward map of only that final input cannot distinguish them. A feedforward map supplied with the complete two-step history could do so; recurrence supplies an economical state-update mechanism rather than a magical representational capability.

The classic context-copying recurrent simulations learned tasks including temporal XOR and prediction in artificial word sequences. Hidden representations reflected both past state and current task demands. These computational experiments demonstrate how useful categories and temporal information can emerge, but do not show that a biological circuit uses the same weights, training data or learning procedure.^[ref]

Recurrent learning is also harder to explain biologically. Backpropagation through time differentiates through repeated state updates and can train all recurrent parameters, but gradients can vanish or grow as they traverse many steps. Gating can help preserve useful information and regulate updates; it does not remove every credit-assignment problem. Activity and stability depend on the operating regime. Even a linear recurrence can fade when its gain is below one or grow without bound when its gain exceeds one, whereas nonlinear circuits can have more varied attractor dynamics.

Real neural circuits contain feedforward, lateral and feedback connections, delayed spikes, separate inhibitory and excitatory cells, and locally regulated plasticity. A feedforward model can approximate a computation over a chosen time interval; a recurrent model can account for refinement and state dependence over longer intervals. Neither is a complete cell-level description. Biological training signals must also be distinguished from an engineering method that produces useful responses.

Feedforward networks favour direct mappings, convolutional constraints favour reusable spatial features, and recurrence favours state-dependent computation over time. Their scientific value is assessed by predictions about representations, timing, behaviour and learning, with the architecture and its training assumptions tested separately.

Sensory genetics shows that evolution can change receptor molecules, their expression and neural organization, and that the relation between genes and perception is neither one-to-one nor confined to protein-coding mutations. Genetic architecture concerns the number, effects, interactions and arrangement of contributing variants in a specified population and environment. Identifying a receptor is one achievement; explaining variation in sensory performance, its development and its evolutionary consequences requires several additional steps.

A foundational result is that a large sensory repertoire can arise by gene duplication followed by divergence. Cloning of human visual-pigment genes revealed related but distinct short-, medium- and long-wavelength-sensitive pigments, with particularly close sequence similarity between the latter two. ^[ref] This links colour vision to a history of related opsins, rather than requiring a wholly new biochemical mechanism for every sensitivity. However, additional gene copies need not produce additional functional sensory channels: copies can be lost, expressed together, or differ too little in sensitivity to support useful discrimination.

The distinction follows from the physics of receptor stimulation. In a simplified model, input to receptor class $i$ is proportional to:where $I$ is the light spectrum reaching the eye and $S_i$ its effective spectral sensitivity. Different spectra can produce the same $R_i$, so a single receptor cannot independently determine wavelength and intensity. Comparing multiple receptor classes allows more information, but the animal also needs suitable neural circuits, a decision process and an appropriate task. Thus an opsin sequence is not a complete description of a perceived colour, and a colour vision test must vary brightness to prevent intensity from supplying the answer.

Regulatory differences can be as important as changes in the receptor sequence. Crosses between cichlid species expressing different opsin combinations identified relatively few regions contributing to expression differences, including both cis-regulatory elements and trans-regulation. ^[ref] The same inherited receptor repertoire can therefore support different visual sensitivities through deployment of different members. A trans-acting regulator can coordinate several genes, whereas a local regulatory change can alter a more restricted expression pattern. These are mechanistic possibilities, not a rule that cis changes are always specific or trans changes always costly. Developmental stage and light exposure also matter, so comparisons need common rearing, matched ages and measurements beyond a single transcript sample.

Experimental addition of a receptor can test whether downstream processing can use the new input. Adding a human long-wavelength opsin to adult dichromatic Saimiri sciureus produced new colour-discrimination performance. ^[ref] The result challenges the assumption that a new sensory channel always requires a matching circuit assembled only in early development. It does not imply that every sensory deficit can be corrected this way or that the monkeys' subjective experience became identical to a human's. Stable pre-intervention performance, tests across chromatic directions and controls against brightness cues make the inference stronger than expression of the introduced gene alone.

Olfaction reveals a related architecture at a larger receptor repertoire. Molecular cloning identified a very large family of candidate olfactory receptor genes with expression in olfactory tissue. ^[ref] This supplies diversity of molecular recognition, but it does not imply one gene for every odour or one receptor for every behavioural response. Odour mixtures and concentrations can engage overlapping inputs, and the same sensory input can acquire different significance through context and learning. Establishing receptor function requires physiological responses to defined stimuli, appropriate expression controls and comparison with cells lacking the receptor.

genes also help organize the neural map receiving those inputs. Labelling mouse neurons expressing a particular receptor showed convergence at characteristic locations in the olfactory bulb. Receptor-swap experiments changed targeting, indicating a contribution of receptor identity to axon guidance. ^[ref] The receptor was not the sole determinant: the resulting projection was not simply an arbitrary map wholly specified by the substituted molecule. This connects molecular recognition with neural development and illustrates pleiotropy, while leaving room for other guidance mechanisms and activity-dependent refinement. A gene can influence the wiring through which its own sensory signal will be interpreted.

Sensory loss likewise need not mean simple destruction of a receptor gene. In cave populations of Astyanax mexicanus, increased embryonic Hedgehog signalling was associated with lens apoptosis and arrested eye development; experimental overexpression in surface fish reproduced important elements of the phenotype. ^[ref] Manipulating a developmental pathway therefore provides causal evidence beyond observing an association between cave residence and blindness. It does not identify every naturally selected causal variant, prove that eye reduction was directly favoured rather than indirectly associated with other traits, or establish the same mechanism in all cave populations. Developmental competence and the adult phenotype can differ substantially.

These findings also connect sensory genetics with speciation. In Lake Victoria cichlids, differences in the light environment, visual pigments, mating signals and preference were associated with reproductive differentiation. ^[ref] The connection is that selection on obtaining information can change the relative effectiveness of signals; it is not a claim that every opsin variant is an isolating barrier. One must separately establish expression, sensitivity, preference and reproductive consequences.

Several general methodological lessons follow. A quantitative trait locus is an interval associated with variation, not automatically the causal gene. A large detected effect does not exclude many small effects, and laboratory crosses may sample only a subset of natural variants. Fine mapping, allele replacement, reciprocal genetic backgrounds, a genetic rescue experiment and neural or behavioural measurements can connect successive causal steps. Pleiotropy, genetic linkage and environmental covariance require different explanations even when all produce correlated sensory traits. Conversely, absence of segregating variation in one laboratory stock says little about all evolutionary routes available to a lineage.

Time-dependent reorganization of the neural circuits supporting retention and retrieval, often involving interactions between the hippocampus and cerebral cortex. Whether a particular detailed episode becomes independent of the hippocampus is a further empirical question.