Christopher Stanton appeals a judgment of the district court upholding the Social Security Commissioner's denial of his applications for child insurance benefits and supplemental security income. An administrative law judge, relying on testimony from a vocational expert, concluded that Stanton's limitations did not prevent him from performing certain jobs in the national economy. But because the administrative law judge did not elicit a reasonable explanation to resolve an apparent conflict between testimony from the vocational expert and the Dictionary of Occupational Titles, we reverse and remand for further proceedings.

I.

In 2013, Stanton applied for child's insurance benefits and supplemental security income under Title II and Title XVI of the Social Security Act, 42 U.S.C. §§ 402 , 1382, based on a bipolar disorder with an onset date of March 1, 2010. After the Social Security Administration denied Stanton's applications initially and on reconsideration, Stanton received a hearing before an administrative law judge in November 2014.

At the hearing, the ALJ heard testimony from a vocational expert. In her hypothetical to the vocational expert, the ALJ asked if there were jobs in the national economy that a person could perform with the following limitations:

no exertional limitations; however, he is limited to unskilled work. Able to understand, retain and carry out simple one to two step instructions, can make simple work related decisions (INAUDIBLE) or judgment retained in the workplace. Interpersonal contact is incidental and specifically limited contact with the public. Supervision is simple, direct and concrete.

The vocational expert responded that such a person could work as a hospital or industrial cleaner. The ALJ then asked the expert: "[I]s your testimony consistent with the information in the [Dictionary of Occupational Titles] and contained in the publication or is it based on your knowledge, education, training and experience?" The vocational expert responded: "Your Honor, it is consistent with the DOT and my experience."

Applying the Commissioner's familiar five-step process, the ALJ then found that Stanton was not disabled. See Bowen v. Yuckert , 482 U.S. 137 , 140-42, 107 S.Ct. 2287 , 96 L.Ed.2d 119 (1987) ; 20 C.F.R. §§ 404.1520 , 416.920 (2012). The ALJ determined at step one that Stanton had not engaged in substantial gainful activity since March 1, 2010. At step two, the ALJ found that Stanton had three severe impairments: attention-deficit hyperactivity disorder, bipolar disorder, and anxiety. But at step three, the ALJ concluded that these impairments did not meet or medically equal the severity of one of the impairments listed in 20 C.F.R. pt. 404, subpt. P, app. 1 (2014). The ALJ then determined Stanton's residual functional capacity for purposes of steps four and five. She found that Stanton had the capacity

to perform a full range of work at all exertional levels but with the following nonexertional limitations: the claimant is limited to unskilled work. The claimant is able to understand, retain and carry out simple one- to two-step instructions. The claimant can make simple work-related decisions, perform work where the complexity of tasks is learned and performed by rote with few variables and little judgment, and can work with few changes in the workplace. He can work where interpersonal contact is incidental to the work performed and where he has limited contact with the public. In addition, he is able to work where the supervision is simple, direct and concrete.

At step four, the ALJ found that Stanton had no past relevant work and that transferability of job skills was therefore not an issue.

At step five, the ALJ determined that the vocational expert's testimony "is consistent with the information contained in the Dictionary of Occupational Titles." Based on that testimony, the ALJ found that Stanton could perform jobs that exist in significant numbers in the national economy-specifically, work as a hospital or industrial cleaner. The ALJ therefore concluded that Stanton was not disabled within the meaning of the Social Security Act and was not eligible for child insurance benefits or supplemental security income. The Appeals Council denied review, and the district court upheld the ALJ's decision.

Stanton appeals, arguing that the ALJ improperly relied on the testimony of the vocational expert at step five, because the ALJ left unresolved an apparent conflict between the vocational expert's testimony and the Dictionary of Occupational Titles. "We review the district court's judgment de novo , considering evidence that both supports and detracts from the ALJ's conclusion." Vance v. Berryhill , 860 F.3d 1114 , 1117 (8th Cir. 2017). "We will affirm if substantial evidence on the record as a whole supports the Commissioner's determination." Id. The substantial evidence threshold is less than a preponderance, but enough that a reasonable mind would find it adequate to support the ALJ's decision. Gonzales v. Barnhart , 465 F.3d 890 , 894 (8th Cir. 2006).

II.

The Dictionary of Occupational Titles provides "standardized occupational information" by listing the functional requirements for a number of jobs available in the national economy. 1 Dictionary of Occupational Titles , at xv (4th rev. ed. 1991). The reference book explains, however, that the job characteristics for each position "reflect[ ] jobs as they have been found to occur, but ... may not coincide in every respect with the content of jobs as performed in particular establishments or at certain localities." Id. at xiii. Thus, "not all of the jobs in every category have requirements identical to or as rigorous as those listed in the [Dictionary]." Wheeler v. Apfel , 224 F.3d 891 , 897 (8th Cir. 2000). The reference book "gives the approximate maximum requirements for each position." Jones v. Chater , 72 F.3d 81 , 82 (8th Cir. 1995). As a result, an ALJ in some instances may rely on testimony from a vocational expert that conflicts with the job requirements listed in the Dictionary-for example, if the expert limits his testimony to jobs within a particular Dictionary description that require less than the listed maximum functional requirements. See Thomas v. Berryhill , 881 F.3d 672 , 677-78 (8th Cir. 2018).

This court, however, long has held that before an ALJ can rely on a vocational expert's testimony that appears to conflict with a Dictionary listing, the ALJ must identify and resolve the conflict. Otherwise, the vocational expert's testimony is not substantial evidence to support a denial of benefits. See Porch v. Chater , 115 F.3d 567 , 572 (8th Cir. 1997) ; Montgomery v. Chater , 69 F.3d 273 , 276 (8th Cir. 1995). In 2000, the Commissioner adopted this rule as its own in Social Security Ruling 00-4p, and we have continued to apply it. Whether the vocational expert's testimony is substantial evidence in support of the ALJ's decision thus depends on whether the expert's testimony appears to conflict with the Dictionary, and if so, whether the ALJ resolved the conflict.

The vocational expert's testimony in this case appears to conflict with the Dictionary. According to the Dictionary, the job identified by the vocational expert-hospital or industrial cleaner-requires a worker to employ Level 2 Reasoning. 1 Dictionary of Occupational Titles 381.687-018, at 282. Level 2 Reasoning, in turn, requires a worker to "[a]pply commonsense understanding to carry out detailed but uninvolved written or oral instructions" and to "[d]eal with problems involving a few concrete variables in or from standardized situations." 2 Dictionary of Occupational Titles 1011. Level 2 Reasoning is a step above Level 1 Reasoning, which requires a worker to "[a]pply commonsense understanding to carry out one- or two-step instructions" and to "[d]eal with standardized situations with occasional or no variables in or from those situations encountered on the job." Id.

The hypothetical posed by the ALJ to the vocational expert and the ALJ's final residual functional capacity determination both limit Stanton to Level 1 Reasoning. At the hearing, the ALJ's hypothetical limited Stanton to the ability "to understand, retain and carry out simple one to two step instructions." Similarly, the ALJ ultimately found that Stanton's residual functional capacity was limited to the ability "to understand, retain and carry out simple one- to two-step instructions." These statements correspond directly to language used by the Dictionary to describe Level 1 Reasoning: "Apply commonsense understanding to carry out one- to two-step instructions." Id. ; see also Thomas , 881 F.3d at 677 . Thus, when the vocational expert testified that someone with Stanton's limitations could perform the job of hospital or industrial cleaner, he testified that someone limited to Level 1 Reasoning could perform a job that the Dictionary describes as having an approximate maximum requirement of Level 2 Reasoning.

The Commissioner, however, urges that no conflict exists between the vocational expert's testimony and the Dictionary. Relying on Moore v. Astrue , 623 F.3d 599 (8th Cir. 2010), the Commissioner observes that a claimant limited to "carrying out simple job instructions" and to performing "simple, routine and repetitive work activity at the unskilled task level" may be able to perform work requiring Level 2 Reasoning. Id. at 604 . But Moore emphasized that the ALJ decision at issue did not limit job instructions " 'to simple one- or two-step instructions' or otherwise indicate that Moore could perform only occupations at a [Dictionary] Level 1 reasoning level." Id. The ALJ here did limit job instructions to one- to two-step instructions, and she thereby indicated that Stanton could perform only occupations requiring Level 1 Reasoning.

At oral argument, the Commissioner presented a new reason why no apparent conflict exists between the vocational expert's testimony and the Dictionary. We typically do not consider theories raised for the first time at oral argument, but we are not persuaded by this one in any event. The Commissioner argued that the ALJ's residual functional capacity finding "overlapped" Level 1 and Level 2 Reasoning, because the finding adopted the "instructions" portion of Level 1 Reasoning and the "variables" portion Level 2 Reasoning. In other words, although the ALJ limited Stanton to carrying out "simple one- to two-step instructions," which corresponds to Level 1 Reasoning, the ALJ also found that Stanton could "perform work where the complexity of tasks is learned and performed by rote with few variables and little judgment," which the Commissioner argues corresponds to Level 2 Reasoning. The Commissioner speculates that the quoted statement in the ALJ's residual functional capacity finding corresponds to the "(INAUDIBLE)" portion of the hypothetical that the ALJ posed to the vocational expert.

Even if true, this overlap does not remove the conflict. At a minimum, the ALJ's limitation to one- to two-step instructions connotes that Stanton cannot perform jobs that would require him to understand and carry out "detailed but uninvolved written or oral instructions" under Level 2 Reasoning. 2 Dictionary of Occupational Titles , 1011. The Dictionary, however, describes the job of hospital or industrial cleaner as having an approximate maximum requirement of Level 2 Reasoning, including the ability to "[a]pply commonsense understanding to carry out detailed but uninvolved written or oral instructions." 1 Dictionary of Occupational Titles 381.687-018, at 282. Thus, according to the Dictionary, the job of hospital or industrial cleaner may require a reasoning ability greater than Stanton's. An apparent conflict remains between the vocational expert's testimony and the Dictionary.

The ALJ did not adequately address this apparent conflict. The ALJ asked the vocational expert whether his testimony was consistent with the Dictionary, but an ALJ must "do more than have the expert affirm that his testimony was consistent with the [Dictionary]." Thomas , 881 F.3d at 678 . Instead, the ALJ must "elicit from the expert an opinion whether there is a 'reasonable explanation' for the conflict and determine whether the expert's testimony warranted reliance despite the conflicting information in the [Dictionary]." Id . (quoting Moore v. Colvin , 769 F.3d 987 , 989-90 (8th Cir. 2014) ). The expert's statement that his opinion was "consistent with the DOT and [his] experience" was not a sufficient explanation, because it did not address whether or why the expert's experience provided a basis to overcome an apparent conflict with the Dictionary. The expert might have learned from experience that the jobs of hospital or industrial cleaner do not require Level 2 Reasoning, but he did not say so. The record also admits of the possibility that the expert simply did not recognize the apparent conflict between his testimony and the Dictionary. Further elaboration was required.

This case is comparable to the recent appeal in Thomas . There, the ALJ posed a hypothetical to the vocational expert that limited the claimant to "1 to 2 step tasks," and the vocational expert identified a job that involved Level 3 Reasoning according to the Dictionary. Id. at 676-77. The court recognized that the ALJ's reference to "1 to 2 step tasks" corresponded to Level 1 Reasoning, creating an apparent conflict between the vocational expert's testimony and the Dictionary. Id. at 677. Because neither party raised the conflict at the hearing, "the expert was not asked to explain why his testimony seemed to differ from the information in the [Dictionary]." Id. The court therefore held that the expert's testimony did not constitute substantial evidence on which the ALJ could rely. Id. at 678. The same inadequacy of explanation here means that the vocational expert's testimony was insufficient to support the ALJ's decision.

* * *

For these reasons, the judgment of the district court is reversed, and the case is remanded with instructions to return the case to the agency for further proceedings.