Inventor Testimony and Corroboration After the New SMED Guidance
Executive Summary: The USPTO’s September 29, 2026 guidance on Subject Matter Eligibility Declarations (SMEDs) provides an opportunity to distinguish legally required corroboration from the factual support that makes inventor testimony persuasive. Drawing on Supreme Court and Federal Circuit decisions and USPTO examination guidance, this article develops a framework that evaluates each assertion according to its purpose, procedural setting, factual foundation, and connection to the claims. Inventors may submit SMEDs, and their interest in the application does not categorically disqualify their testimony; corroboration requirements nevertheless remain applicable where the governing doctrine requires independent confirmation. The article explains how patents, journal articles, technical standards, testing, and contemporaneous records can substantiate technical explanations, establish meaningful comparisons, and support claimed improvements. It also distinguishes evidence measuring a disclosed improvement from information improperly supplying a mechanism missing from the original disclosure. Applying these principles to technological improvement, mental processes, conventionality, and historical invention assertions, the article offers practical guidance for preparing declarations and preserving the examination record. Its central recommendation is to connect each material assertion to identifiable support and the relevant claim language, while addressing the actual evidentiary weakness rather than treating inventor status as dispositive.
Introduction
An inventor may be best positioned to explain how a claimed technology works. That knowledge can be valuable when a subject matter eligibility rejection rests on an incomplete account of the invention. The USPTO’s September 29, 2026 guidance on Subject Matter Eligibility Declarations, or SMEDs, encourages applicants to develop that factual record while emphasizing objective evidence and disinterested sources. When must an inventor’s testimony be corroborated, and when does independent support simply make an otherwise competent explanation more persuasive?
The answer depends on the proposition the testimony is offered to establish and the procedural setting in which it is offered. The guidance creates an occasion to distinguish legally required corroboration from the factual support needed to make technical testimony persuasive. This article offers a prospective framework for making that distinction. It neither attributes a change in corroboration doctrine to the memorandum nor assumes that examiners routinely require independent experts. Its central proposition is that each assertion should be evaluated according to its purpose, procedural setting, foundation, and connection to the claims, with the witness’s interest receiving appropriate consideration.
What the Guidance Requires
The September memorandum supplements the April 30, 2026 applicant guidance, leaves the December 4, 2025 examiner memorandum unchanged, and expressly announces no new practice or procedure. SMED submission remains optional, and the stated best practices are recommendations rather than additional requirements. Memorandum from John A. Squires, Under Sec’y of Com. for Intell. Prop. & Dir., U.S. Pat. & Trademark Off., to All Patent Applicants & Patent Practitioners, Subject Matter Eligibility Declarations (SMEDs) Under 37 CFR 1.132—Flowchart for Analysis and Best Practices 2–4 (Sept. 29, 2026) [hereinafter September Memorandum]. The practical task is to work within existing declaration and examination rules while developing evidence directed to the actual eligibility dispute.
The unchanged December examiner memorandum expressly permits an inventor with knowledge of the asserted facts to sign a declaration. Its neural-network example also contemplates an inventor presenting objective comparative-testing results. Inventor status therefore does not disqualify the witness from supplying eligibility evidence; the remaining questions concern the assertion’s foundation, relevance, and weight. Memorandum from John A. Squires, Under Sec’y of Com. for Intell. Prop. & Dir., U.S. Pat. & Trademark Off., to Patent Examining Corps, Subject Matter Eligibility Declarations 2, 5 (Dec. 4, 2025) [hereinafter December Examiner Memorandum].
The accompanying flowchart separates procedural consideration from evidentiary weight. Part 1 addresses timeliness, the required declaration statements, and the declarant’s signature and knowledge of the asserted facts. Part 2 asks whether the submission presents objective evidence, whether that evidence has a sufficient connection to the claim, and whether it supplies information required in the original disclosure. Both parts are reproduced after this article. U.S. Pat. & Trademark Off., Evaluating Subject Matter Eligibility Declarations (SMEDs) pts. 1–2 (2026) [hereinafter SMED Flowchart].
The destination of Part 2’s adverse branches matters. A lack of objective evidence, an insufficient claim connection, or an attempt to supply required original disclosure leads to lower or no probative value. Those branches do not return the declaration to Part 1’s finding of formal insufficiency. The examiner proceeds to weigh the evidence of record and explain the eligibility determination. This structure does not rehabilitate missing disclosure or give every assertion meaningful weight. It recognizes that evidence properly before the examiner may nevertheless deserve little weight for an identified substantive reason. SMED Flowchart, supra, pt. 2.
The memorandum similarly treats witness interest as relevant to persuasiveness. An applicant’s declaration concerning the advantages of the claimed invention may not be disregarded merely because of its source, although a disinterested person’s testimony may carry greater weight. Publications from neutral sources may provide a factual basis for either witness’s explanation. September Memorandum, supra, at 3–4. That guidance invites scrutiny of the evidence and its foundation; it does not establish an independent-expert prerequisite.
Eligibility doctrine explains why technical evidence can matter. Although eligibility is ultimately a legal question, it may involve underlying factual issues, including whether a claim element or combination was well-understood, routine, and conventional to a skilled artisan. Berkheimer v. HP Inc., 881 F.3d 1360, 1368–69 (Fed. Cir. 2018). The court also cautioned that not every eligibility dispute presents a genuine dispute over material facts. Id. at 1368. A declaration may illuminate a material factual premise without requiring competing experts in every case or compelling a finding of eligibility. The corroboration cases address how particular testimony should be evaluated; eligibility doctrine identifies which factual propositions may affect the legal result. Berkheimer’s litigation discussion does not displace the examination standard addressed below.
When Corroboration Is Legally Required
The Supreme Court’s historical cases explain the reliability concerns behind corroboration. Recollections of old inventive activity can be shaped by interest, mistaken memory, or an honest failure to recall the feature that distinguishes the patented invention. See The Barbed Wire Patent, 143 U.S. 275, 284–85, 289 (1892); Deering v. Winona Harvester Works, 155 U.S. 286, 300–01 (1894). Unsupported accounts of invention many years earlier therefore warrant particular caution. See T.H. Symington Co. v. Nat’l Malleable Castings Co., 250 U.S. 383, 386 (1919).
The needed support can take different forms. In Smith v. Hall, the Court found oral prior-use testimony insufficient by itself but examined contemporaneous publications and an abandoned application and its prosecution record as corroboration. 301 U.S. 216, 222–27 (1937). Corona Cord Tire likewise recognized that a single witness’s account could carry weight when supported by circumstances and records. Corona Cord Tire Co. v. Dovan Chem. Corp., 276 U.S. 358, 381–83 (1928). These cases establish the importance of reliable proof of historical activity without invariably demanding another eyewitness.
In contested proceedings, the Federal Circuit has applied corroboration to testimony establishing prior conception or derivation, earlier invention offered to overcome prior art, and omitted inventorship. Price v. Symsek, 988 F.2d 1187, 1194–96 (Fed. Cir. 1993); Mahurkar v. C.R. Bard, Inc., 79 F.3d 1572, 1577–78 (Fed. Cir. 1996); Blue Gentian, LLC v. Tristar Prods., Inc., 70 F.4th 1351, 1357, 1362–63 (Fed. Cir. 2023). The patentee’s defensive posture does not eliminate the requirement when the assertion concerns an earlier invention date. Conversely, financial disinterest does not eliminate corroboration when testimony alone is offered to establish invalidating activity. Finnigan Corp. v. Int’l Trade Comm’n, 180 F.3d 1354, 1366–70 (Fed. Cir. 1999); see also Woodland Tr. v. Flowertree Nursery, Inc., 148 F.3d 1368, 1371–73 (Fed. Cir. 1998).
Courts evaluate corroboration under a rule of reason, considering the evidence collectively. Corroborating material need not independently establish every claim limitation, but the showing cannot depend solely on the inventor’s own assertions. Fleming v. Escort Inc., 774 F.3d 1371, 1377 (Fed. Cir. 2014); Medichem, S.A. v. Rolabo, S.L., 437 F.3d 1157, 1170–71 (Fed. Cir. 2006). The support must also match the proposition: evidence of possession of a chemical did not necessarily establish that the inventors had made it using the claimed process. Medichem, 437 F.3d at 1172–73.
Procedural setting can affect the corroboration requirement itself. The MPEP distinguishes Rule 131 antedating practice from interference practice, retaining language permitting an applicant to rely on the applicant’s own declaration. Manual of Patent Examining Procedure § 715.07(III) (9th ed., rev. 01.2024, Nov. 2024) [hereinafter MPEP] (citing Ex parte Hook, 102 U.S.P.Q. 130 (Bd. App. 1953)). In the AIA implementation rulemaking concerning Rules 130 and 131, the Office likewise declined to impose a per se corroboration requirement in ex parte examination and reexamination, describing its necessity as dependent on the particular facts. Changes To Implement the First Inventor To File Provisions of the Leahy-Smith America Invents Act, 78 Fed. Reg. 11,024, 11,044 (Feb. 14, 2013) (response to comment 34).
Those administrative formulations do not displace binding judicial authority. In re NTP required corroboration of an asserted earlier invention date in reexamination and rejected circular reliance on the inventors’ testimony to establish the date of a document offered to corroborate them. In re NTP, Inc., 654 F.3d 1279, 1291–92 (Fed. Cir. 2011). In re Steed required sufficiently specific documentary support when an applicant sought to antedate a reference. 802 F.3d 1311, 1316 (Fed. Cir. 2015). The MPEP itself includes both decisions alongside its older formulation. MPEP § 715.07(I), (III). The analysis therefore requires the authorities governing the assertion’s particular use; neither historical subject matter nor an ex parte setting alone resolves it.
Corroboration and the burden of proof remain separate questions. The modern standard for an invalidity defense under 35 U.S.C. § 282 is clear and convincing evidence. Microsoft Corp. v. i4i Ltd. P’ship, 564 U.S. 91, 95, 100–05 (2011). Older references to proof beyond a reasonable doubt do not supply today’s standard. See Price, 988 F.2d at 1191–94. Nor should the litigation burden be imported into ordinary examination merely because some corroboration decisions arose in validity disputes.
Why Inventor Testimony Is Not Categorically Deficient
In re Reuter illustrates why the nature of the assertion matters. The same witness offered an expert account of the state of the art and assertions concerning his own earlier reduction to practice. The court stated that expert opinion in an affidavit need not be corroborated, distinguished assertions of earlier invention, and gave no weight to an opinion on the ultimate legal issue. 670 F.2d 1015, 1021–23 (C.C.P.A. 1981). Farrenkopf later preserved Reuter’s statement concerning the treatment of affidavit evidence. In re Farrenkopf, 713 F.2d 714, 720 n.6 (Fed. Cir. 1983). But Thomson characterized Reuter’s statement withholding weight from an uncorroborated assertion of an earlier anticipating invention as dictum. Thomson, S.A. v. Quixote Corp., 166 F.3d 1172, 1176 n.5 (Fed. Cir. 1999). Reuter thus illustrates assertion-specific analysis; the authorities directly addressing earlier invention supply the mandatory-corroboration rules.
The Federal Circuit’s i4i decision supplies a related limit. The inventors testified that an earlier software product lacked a limitation of the patented invention. The court rejected an argument that this testimony required corroboration, distinguishing it from testimony establishing an earlier invention date. i4i Ltd. P’ship v. Microsoft Corp., 598 F.3d 831, 847 (Fed. Cir. 2010). That distinction neither exempts every inventor defending validity nor supports a categorical rule that all inventor testimony requires independent confirmation. The Supreme Court’s later decision addressed the invalidity burden, not a separate corroboration holding. See Microsoft, 564 U.S. at 95.
EmeraChem shows why rejecting a categorical rule does not excuse a deficient declaration. An inventor asserted that relevant portions of a reference originated with the same inventive entity as the challenged patent, but supplied no supporting context, explanation, or evidence. The court found that bare assertion inadequate while expressly declining to require corroboration of an inventor declaration in every case or contemporaneous documentary evidence in all circumstances. EmeraChem Holdings, LLC v. Volkswagen Grp. of Am., Inc., 859 F.3d 1341, 1345–48 (Fed. Cir. 2017).
Four analytical questions help explain these distinctions. Mandatory corroboration asks whether the doctrine governing the particular assertion in its procedural setting requires independent confirmation. Factual foundation asks what observations, technical reasoning, methodology, or records support the assertion. Objectivity asks what makes the asserted fact assessable beyond the declarant’s assurance that it is true. Disinterest asks what relationship or incentive might affect the weight assigned to the account. These are explanatory categories, not four additional formal tests. They prevent inventor status, interest, lack of corroboration, and lack of objective support from becoming interchangeable labels.
Documentary evidence makes the distinction concrete. In an America Invents Act derivation proceeding, Global Health Solutions upheld the finding that contemporaneous emails and surrounding evidence adequately corroborated independent conception. Global Health Sols. LLC v. Selner, No. 2023-2009, slip op. at 13–15 (Fed. Cir. Aug. 26, 2025). The court held that the emails, whose authenticity was unchallenged, did not themselves require independent corroboration. It separately reasoned that, in any event, the server-generated metadata was independent of the inventor’s authorship. Id. at 15. Those are distinct points: the decision does not demand another independent source for every authentic documentary exhibit or make metadata a universal prerequisite. The practical inference is that authorship is only one consideration; timing, authenticity, content, and the surrounding record also matter.
An outside expert does not necessarily repair an inadequate factual foundation. If the expert merely repeats an inventor’s unsupported claim of improved performance, the new signature adds little to the methodology or claim connection. Conversely, an inventor may explain a technical proposition through records and reasoning that permit meaningful assessment. Witness selection should follow the knowledge needed to establish the relevant facts.
The strongest objection to relying on inventor testimony deserves a direct response. Interested witnesses may overstate technological significance, select favorable comparisons, or describe an implementation more narrowly than the claims. These risks justify searching examination of the evidence. The response should identify the actual weakness: an unsuitable baseline, missing test conditions, an unsupported assumption, an unclaimed feature, or a reason to doubt credibility. Treating inventor status as dispositive would obscure those more useful inquiries. See September Memorandum, supra, at 3–4; MPEP § 716.01(c).
Application to Specific SMED Assertions
These applications of the corroboration cases to SMED practice are doctrinal inferences; the decisions did not themselves adjudicate SMEDs. Corroborating evidence can strengthen an inventor’s SMED even when independent confirmation is not a legal prerequisite to the particular assertion. Other patents, published applications, journal articles, textbooks, technical standards, and contemporaneous manuals can give the examiner a basis for assessing the inventor’s explanation. The September memorandum expressly illustrates this approach: a trade article describing technological limitations known at filing may support an expert’s explanation that a skilled artisan would recognize the disclosed improvement and that the claims reflect it. September Memorandum, supra, at 3–4. Properly submitted publications should be considered to the extent they are probative of the disputed facts. MPEP § 716.
The supporting source should be tied to a specific proposition. Consider an inventor who explains that a disclosed data structure avoids reconstructing stored data during a claimed operation, reports fewer memory accesses, describes industry practices, and asserts an earlier conception date. A journal article might substantiate the cost of repeated reconstruction; another patent might document the architecture used for comparison; test records might support the measured reduction. Those sources would serve different purposes, and none necessarily establishes when this inventor conceived the claimed subject matter.
Other patents can be particularly useful when their specifications identify a recognized technical problem, explain how an earlier architecture operates, or describe the limitations of a particular approach. The inventor should identify the relevant paragraphs, columns, or figures and explain their technical relationship to the declaration’s assertion. In the data-structure example, a cited patent might show that an earlier system reconstructs an entire object before updating one field. The inventor could then explain, with citations to the application and the claim language, how the claimed structure permits the update without that reconstruction. The useful evidence is the patent’s technical teaching and the reasoned comparison. The fact that another patent issued does not itself establish that the pending claims are eligible. See In re Wertheim, 541 F.2d 257, 264 (C.C.P.A. 1976).
Journal articles can provide complementary support. A review article may document a recognized bottleneck or the approaches used in a field, while an experimental paper may provide measurements supporting a technical principle or a proposed baseline. Peer review is relevant to evaluating a source, but it does not establish that the article’s methods, assumptions, or results apply to the claimed invention. The declaration should explain that connection. A paper describing reconstruction overhead, for example, can support the premise that avoiding reconstruction has technical significance; it does not by itself prove the magnitude of the applicant’s asserted performance gain.
For measured performance, the declaration should identify the tested implementation, baseline, workload, conditions, and records supporting the result. A journal article or industry benchmark can help justify the comparison, while laboratory records, instrument logs, source-code versions, or a third-party test report may support what was actually measured. Differences between the published benchmark and the applicant’s implementation should be explained. A faster commercial product proves little about the claimed improvement if an unclaimed component accounts for the difference. Testing nevertheless is not a universal prerequisite: some propositions can be persuasively explained through the disclosed architecture and established technical principles.
For the mental-process inquiry, a patent, standard, or technical manual may help explain the machine operations required by a particular claim limitation. The inventor should connect the reference’s description to the actual claimed operation and explain why that operation cannot practically be performed in the human mind. Evidence that a commercial implementation is fast or processes large volumes of information is less useful when the claim does not require the features producing those characteristics. The memorandum identifies the mental-process inquiry as an appropriate subject for a SMED, but defeating that characterization does not necessarily resolve a separate mathematical-concept rationale. September Memorandum, supra, at 4; MPEP §§ 2106.04(a)(2), 2106.07(b).
For technological improvement, external literature should support the explanation of how the claimed mechanism changes computer or other technical operation. The USPTO’s guidance following Ex parte Desjardins focuses on whether the specification discloses enough for a skilled person to recognize the improvement and whether the claims reflect the features producing it. The claims need not expressly recite the resulting benefit. Memorandum from Charles Kim, Deputy Comm’r for Pats., U.S. Pat. & Trademark Off., to Patent Examining Corps, Advance Notice of Change to the MPEP in Light of Ex Parte Desjardins 2–4 (Dec. 5, 2025). Publications can help establish the technical context in which a skilled artisan reads the disclosure. They cannot supply a missing mechanism required in the original disclosure or import unclaimed features into the eligibility analysis. SMED Flowchart, supra, pt. 2.
The absence of supporting measurements from the original application is not necessarily the same defect as missing disclosure of the technical mechanism those measurements explain. In the December memorandum’s neural-network example, the specification describes the claimed architecture and asserts superior performance but omits the specific performance improvements. An inventor’s SMED supplies comparative regression-test results, subject to the original-disclosure and claim-nexus requirements. December Examiner Memorandum, supra, at 3, 5. Later evidence may substantiate a disclosed improvement; it cannot add architecture needed to make that improvement recognizable.
For conventionality, contemporary surveys, textbooks, standards, and technical literature may substantiate an account of industry practices or unresolved technical difficulties. The declaration should identify the relevant field and period and distinguish the claimed arrangement from its individual components. A single patent’s disclosure of a feature does not establish that the feature was well-understood, routine, and conventional; the record must support that further inference. Berkheimer, 881 F.3d at 1369; MPEP § 2106.07(a)(III)(C). Conversely, the absence of the claimed arrangement from selected publications does not, by itself, prove nonconventionality. This inquiry belongs to Step 2B and should not be imported into Step 2A, Prong Two. MPEP §§ 2106.04(d), 2106.05(d).
Dates and provenance should be addressed according to the proposition being supported. A publication used to establish what skilled artisans knew at filing should have an identified publication date and a demonstrated connection to that period. A later article may illuminate a technical principle, but its later appearance alone does not establish earlier knowledge or practice. Likewise, an inventor-authored patent or paper may be useful evidence without being disinterested evidence. Authorship, timing, methodology, and corroborating records should be evaluated together, as the treatment of electronic records in Global Health Solutions illustrates. See Global Health Sols., No. 2023-2009, slip op. at 13–15.
If the same declaration also asserts earlier invention to antedate a reference under an applicable pre-AIA provision, that assertion requires separate analysis under Rule 131 and its governing authorities. General literature establishing a technical problem does not establish when this inventor possessed the claimed solution. The declaration and supporting records must address the legally relevant acts, dates, and claimed subject matter. NTP, 654 F.3d at 1291–92; Steed, 802 F.3d at 1316. Rule 131(a) is not a general means of overcoming AIA prior art merely by showing earlier conception. 37 C.F.R. § 1.131(d)–(e) (2026); MPEP § 715. A supported technical explanation does not substitute for the required historical showing, and a deficiency in that showing need not defeat every other assertion.
The submission should make the relationship between testimony and exhibits explicit. Identify each source fully, provide the relied-on material with sufficient context, and cite precise passages in the declaration. For each passage, explain the factual proposition it supports and the claim limitation or disclosed mechanism to which it relates. Distinguish what the inventor personally observed, what the publication reports, and what the inventor infers from the two. Used this way, corroborating material supplies an assessable foundation for the inventor’s explanation rather than merely increasing the number of attached references.
Building and Preserving the Examination Record
Preparation should begin with the disputed limitation and the relevant disclosure passage. Counsel can then identify the factual proposition requiring explanation and its procedural use, select a witness with pertinent knowledge, and gather the supporting material. Separating observations, technical inferences, historical assertions, and legal argument makes the declaration easier to evaluate and exposes gaps before filing. Records supporting a historical account should be retained with information bearing on their dates and authenticity; test results should be retained with the conditions and underlying data necessary to understand them.
The memorandum encourages declarations devoted to eligibility, while permitting a single declaration addressing multiple statutory grounds. It also recommends identifying the precise eligibility issue and filing early, potentially before the first action on the merits. September Memorandum, supra, at 3–4. Separate treatment can clarify the analysis, but it does not relax the rules governing entry of evidence late in prosecution.
After final rejection, evidence filed before or on the notice-of-appeal date may be admitted upon a showing of good and sufficient reasons why it is necessary and was not presented earlier. 37 C.F.R. § 1.116(e) (2026). Evidence filed after that date but before the appeal-brief date faces two conjunctive conditions: the examiner must determine both that it overcomes all rejections under appeal and that a showing of good and sufficient reasons for its necessity and earlier omission has been made. Id. § 41.33(d)(1). A declaration addressing only § 101 therefore does not satisfy that entry route while other appealed rejections remain unresolved. The rule permits admission when its conditions are met; it does not make admission automatic.
Once the evidence is entered, the examiner should address its substance. General statements that a declaration lacks technical validity or fails to match the claims require an explanation. MPEP § 716.01. If the concern is corroboration, the response should identify the particular assertion, its procedural setting, and the doctrine said to require independent confirmation. If the concern involves methodology, qualifications, or claim nexus, the applicant should address that deficiency directly. An interview can clarify the disagreement, and the written record should identify the material evidence and explain its significance to the relevant claims. See September Memorandum, supra, at 4; MPEP § 713.04.
Part 2 of the flowchart supports that disciplined exchange. The examiner weighs the record and explains why the evidence, considered with the other evidence, establishes eligibility or is insufficient. SMED Flowchart, supra, pt. 2. A response should therefore connect the declaration’s factual showing to the examiner’s stated rationale, rather than merely insist that the declaration is unrebutted. Where a deficiency can be cured, the applicant should supply the needed explanation or support while an available procedural route remains open.
The governing examination standard is preponderance of the evidence. The flowchart states that rejection is proper only when the claims are more likely than not ineligible and that an evenly balanced record favors the applicant. SMED Flowchart, supra, pt. 2 n.*. That instruction concerns the complete record after the evidence has been weighed. It does not compel acceptance of every unrebutted assertion, eliminate corroboration where required, or resolve other grounds of rejection. Neither the clear-and-convincing litigation burden nor the summary-judgment posture of Berkheimer replaces this examination framework.
The September guidance provides an occasion to make eligibility disputes more precise. The useful question is what a particular assertion establishes, under which procedural rules, and why the record supports it. That inquiry preserves mandatory corroboration where the governing doctrine requires it, gives witness interest its proper significance, and directs applicants toward evidence that resolves the actual technical dispute. An inventor’s explanation should receive the consideration warranted by its foundation and its connection to the claims.
