PER CURIAM.

Jonas H. Kushner, Kushner & Kushner, Fort Myers, FL, for Plaintiff-Appellant.

Richard Vincent Blake, Owen Keegan, Depak Sathy, Brian Seinberg, Selisa McKay Wright, Social Security Administration, Office of the General Counsel, Atlanta, GA, U.S. Attorney Service-Middle District of Florida, U.S. Attorney's Office, Tampa, FL, for Defendant-Appellee.

Before WILLIAM PRYOR and ROSENBAUM, Circuit Judges, and MOORE, District Judge.

PER CURIAM:

Appellant Hans Schink applied for Social Security disability benefits based on various physical impairments and the fact that he suffered from bipolar disorder. The matter proceeded to a hearing before an administrative law judge ("ALJ"), who denied benefits. Schink appealed the decision and the Appeals Council remanded the matter to the ALJ for further proceedings. Schink fared no better the second time the ALJ considered his case.

Following the second denial by the ALJ, Schink again appealed. But this time the Appeals Council affirmed the denial of benefits. Schink then filed a complaint with the district court, which affirmed the decision to deny benefits. Schink now asks us to find that the ALJ erred by (1) discounting his treating physicians’ opinions and (2) concluding that his bipolar disorder was not a severe impairment. Schink also contends that remand to a different ALJ is warranted because of a high risk that the ALJ who considered his claims was biased against him.

After careful review, we conclude that Schink’s claim of bias was forfeited, but we also conclude that the ALJ’s decision contains errors that must be addressed. Specifically, we find that the ALJ failed to articulate good cause for discounting two treating physicians’ opinions, that substantial evidence does not support the finding that Schink’s bipolar disorder was non-severe, and that the ALJ failed to consider Schink’s mental impairments when assessing his residual functional capacity. We therefore affirm in part and reverse in part the order of the district court affirming the denial of benefits, and we remand with instructions to vacate the Commissioner’s decision and to remand to the Commissioner for further proceedings.

I. Facts

A. Background

Schink applied for disability insurance benefits in February 2010, alleging an onset date of October 1, 2004. He claimed disability due to bipolar disorder, type-2 diabetes, and various physical impairments. Schink remained insured through September 30, 2011, so he was required to establish disability on or before this date to be entitled to benefits. As for other relevant characteristics, Schink has a high school education and past relevant work as a car salesman.

B. Medical Evidence

In setting forth a summary of the relevant medical evidence, we focus on only Schink’s mental impairments, since those are at issue in this appeal.

1. Schink’s Initial Treatment

Although Schink produced extensive medical records, we do not discuss every detail relating to Schink’s mental health. Nevertheless, we note that records indicate Schink had a history of bipolar disorder and a family history of depression. In June 2008, when we pick up Schink’s more recent medical history, doctors believed that antidepressant therapy would be beneficial, and Schink began taking Lexapro.

Notes from psychotherapy sessions in the Spring of 2009 indicate that Schink’s speech was pressured, his mood was agitated, his affect was limited, his judgment was poor, and his relationships were isolated. During this timeframe, Schink met regularly with psychotherapist Nicholas Anthony, Ph.D., who diagnosed Schink with bipolar disorder. Dr. Anthony determined that Schink displayed symptoms of aggression, anger, and agitation, as well as scattered concentration. At times, Schink’s condition improved, though he continuously suffered from bipolar disorder. Dr. Anthony also found Schink’s affect to be "blunted" and his energy to be low, and he concluded that Schink had "marginal social and interpersonal involvement."

In 2010, Schink met with other doctors who similarly documented Schink’s chronic mood swings, depression, anger, and anxiety. During this timeframe, Schink revealed that his father had committed suicide by jumping off a bridge, his mother had died at age 48 (and suffered from depression), and his brother had been murdered. Psychiatrist Raymond Johnson, M.D., recorded that Schink was "extremely hyperverbal and angry" as he fantasized about "get[ting] back at people" who upset him. Schink was consistently diagnosed with bipolar disorder and in mid-2010 was assigned a global assessment of functioning ("GAF") score of 55. Id. Dr. Johnson also noted that Schink had intermixed manic and depressive episodes, racing thoughts, and rapid cycling manic and depressive episodes.

2. State Doctors’ Assessment of Schink

Because he filed for disability benefits in February 2010, Schink was referred for a consultative psychological examination with J.L. Bernard, Ph.D., on June 30, 2010. During the examination, Schink reported that he was agitated, felt like he could kill someone, was very depressed, had memory problems, and had passive suicidal thoughts. Dr. Bernard noted that Schink was talkative, but on several occasions, Schink could not offer details on how he spent portions of his life. Schink reported that he discontinued work because he could "no longer deal with people." He further told Dr. Bernard that he spent most of his time watching television, walking the dog, doing very little housework, napping, playing on his computer, and going for drives. And he told Dr. Bernard that he cooked "minimally" and "read once in a while."

Dr. Bernard indicated that Schink’s attitude at the interview was "brusque, arrogant, flippant, and abrasive," with "a harshness and domineering aspect to his personality" and an "irritable" affect, although his mood was stable. The doctor also reported that Schink had decreased memory skills, pressured speech, and felt like "killing people most of the time." Dr. Bernard diagnosed Schink with mood disorder, personality disorder not otherwise specified with cluster B features, problems dealing with the social environment, and occupational limitations. Dr. Bernard assigned Schink a GAF score of 59 and indicated that his prognosis was chronic.

In furtherance of the disability claim, state agency consulting psychologist Anne-Marie Bercik, Ph.D., conducted a review of Schink’s psychiatric medical history on August 30, 2010. She did not meet with Schink in person. Using a checklist and a scale of "mild," "moderate," "marked," and "extreme," Dr. Bercik concluded that Schink had only mild limitations of daily living, maintaining social functioning, and maintaining concentration, persistence, or pace, and had no episodes of decompensation. Dr. Bercik’s overall impression after reviewing Dr. Bernard’s notes was that while Schink had some mental deficits, his impairments were not severe and did not currently meet or equal a mental listing.

3. Additional Treatment by Schink’s Doctors

Schink returned to see Dr. Anthony in the Fall of 2010, at which point he had been taking Klonopin for anxiety and Celexa for depression for approximately two-to-three months. Dr. Anthony completed a formal assessment of Schink and, on a scale that included "mild," "moderate," and "severe," Dr. Anthony concluded that Schink had "moderate" symptoms of loss of interest or pleasure, sleep disturbance, loss of energy, irritable mood, and cognitive impairment. Dr. Anthony found moderate improvement in reduction of agitation and minimal improvement in Schink’s mood, and he assigned a GAF score of 50.

In April 2011, Schink began to see psychiatrist Nelson A. Hernandez, M.D. Schink complained of racing thoughts, depression, poor sleep, and increased anxiety. Dr. Hernandez completed a Mental Status Examination form, which set forth his opinion that Schink’s affect was labile and his anxiety was moderate, and that he exhibited depression with anhedonia. Dr. Hernandez also indicated that Schink’s mood was dysthymic, his recent memory was impaired, his judgment was fair, and his thought organization was circumstantial. Dr. Hernandez diagnosed Schink with bipolar disorder and anxiety disorder and assigned a GAF score of 60. Dr. Hernandez recommended that Schink begin taking Zoloft and referred him for treatment by Dr. Charles Assad.

Schink returned to see Dr. Hernandez twice in September 2011. At the first meeting, Schink reported having fair energy level, less depression and fewer mood swings, but he noted he still had some racing thoughts. At the second meeting, Schink stated that he was feeling better with less agitation and better sleep.

Based on Dr. Hernandez’s recommendation, Schink began mental-health treatment with psychologist Charles Assad, Ph.D., in June 2011. At that time, Dr. Assad noted that Schink was poorly groomed and that he had pressured speech. Dr. Assad also described Schink as having a cooperative attitude, elevated anxiety and depression, and hypomanic affect. Dr. Assad diagnosed Schink with bipolar disorder and depression, and assigned a GAF score of 55. During a follow-up visit later that month, Dr. Assad found that Schink continued to present with similar symptoms. The next month, however, Dr. Assad found Schink’s thought processes were "clearer and more logical" and that he had less pressured speech, but his "bipolar lability [was clearly] continuing." During an appointment in late July 2011, Dr. Assad again noted rapid speech and tangential thought processes. Schink met with Dr. Assad several more times through October 2011. During these visits, Dr. Assad determined Schink had a depressed mood and affect as well as anger and resentfulness.

Schink returned to see Dr. Assad various times from October 2011 through 2012 and 2013, on a biweekly basis. Dr. Assad’s records reflect that Schink’s impulsiveness and irascibility caused him continuing trouble with relationships and interactions with strangers, that he suffered from financial problems, and that he struggled to follow through on scheduling medical appointments and dealing with other logistic issues in his life.

4. Questionnaires Completed by Drs. Assad and Hernandez

On October 11, 2011, Dr. Assad completed a questionnaire concerning Schink’s mental residual functional capacity, in which he assessed Schink’s ability to engage in work-related activities on a day-to-day basis. According to Dr. Assad, Schink had "marked" limitations in his abilities to (1) accept instruction from or respond appropriately to criticism from supervisors or superiors, (2) work in coordination with or in proximity to others without distracting them or exhibiting behavioral extremes, (3) respond appropriately to coworkers or peers, (4) relate to the general public and maintain socially appropriate behavior, (5) maintain attention and concentration for more than brief periods, (6) perform at production levels expected by most employers, (7) respond appropriately to changes in work setting, (8) maintain personal appearance and hygiene, and (9) tolerate customary work pressures. Dr. Assad also opined that Schink had "extreme" limitations in his ability to behave predictably, reliably, and in an emotionally stable manner. Finally, Dr. Assad estimated that Schink had "mild" limitations in other areas. The questionnaire defined the terms "mild," "marked," and "extreme" as used by Dr. Assad.

In late September 2011, Dr. Hernandez filled out a similar questionnaire. He did not indicate any "extreme" limitations, but he reported "marked" limitations in the areas of Schink’s ability to behave in a predictable, reliable, and emotionally stable manner, and in his ability to tolerate customary work pressures. In all other areas, Dr. Hernandez found Schink to have "moderate" limitations. Dr. Hernandez also indicated that if Schink were placed under stress, Schink’s condition would likely deteriorate. Dr. Hernandez based this assessment on the fact that Schink had showed multiple "flare-ups."

5. Schink’s Voluntary Hospitalization

Schink was voluntarily hospitalized for one week at Park Royal Hospital from December 13, 2013, through December 20, 2013. A discharge summary explains that upon admission, Schink was in distress, had mood swings, was depressed, and was placed on supervision every fifteen minutes to ensure his safety. Schink was given lithium, Wellbutrin, and Ativan. The lithium was later replaced with Trileptal, and Schink was started on Abilify. Upon discharge from the hospital, Schink fared better, denying depression, anxiety, or suicidal plans. The discharge summary listed bipolar disorder, type 2, most recent episode depressed, and mood disorder.

C. ALJ, Appeals Council, and District Court Decisions

In late October 2011, Schink appeared before the ALJ for a hearing on his disability claim. On December 30, 2011, the ALJ issued an unfavorable decision. Schink filed a request for review of the ALJ’s decision, and the Appeals Council remanded the claim by Order dated June 18, 2013.

The ALJ held a de novo hearing on January 28, 2014, as a result of the Appeals Council’s remand order. Schink testified at the hearing that two different employers had fired him after less than three days of employment due to his difficulty controlling his anger and the way he spoke to customers. He testified that it was "really hard for [him] to deal with people" because "sometimes they really aggravate[d] [him] very bad." He also stated that he "[didn’t] really cook or anything" and that he no longer drove much because he "g[o]t very, very angry at people driving." At one point, the ALJ remarked that Schink had cried "a couple of times during the hearing" and asked if that was "normal" for him. Schink replied that he was "upset" and "embarrassed" to be at the hearing, that he "want[ed] to be able to do something," and that he felt like he was "falling apart." He added, "I used to be okay. I don’t know what happened to me, you know." The ALJ issued another unfavorable decision on March 16, 2015, concluding Schink was not under a disability within the meaning of the Social Security Act from October 1, 2004 (the alleged date of onset of disability), through September 30, 2011 (the date of last insured).

Although the ALJ determined that Schink suffered from various physical impairments that were severe, he found that Schink’s bipolar disorder was not severe. In making this determination, the ALJ discussed Schink’s treatment with Drs. Anthony, Hernandez, and Assad, as well as the questionnaires regarding Schink’s Mental Residual Functional Capacity completed by Drs. Hernandez and Assad. He also acknowledged the psychological evaluation completed by Dr. Bernard on June 30, 2010.

The ALJ accorded minimal weight to Dr. Hernandez’s and Dr. Assad’s opinions as set forth in their respective questionnaires. He explained that he did so, among other reasons, because the questionnaires used terms—including "mild," "extreme," and "unable to function"—that either did not appear in official forms used by the Social Security Administration or struck the ALJ as vague or ill-defined. As a result, the ALJ deemed the questionnaires ambiguous with respect to both the questions asked and the providers’ responses. He also objected that the questionnaire did not address the category of "Understanding and Memory." The ALJ further accorded minimal weight to the treating doctors’ opinions because he concluded that they were not well-supported by medically acceptable clinical and laboratory diagnostic techniques and were inconsistent with other evidence in the record. He also stated that the doctors provided sporadic treatment and their treatment notes reflected only mild limitations.

In support of his findings, the ALJ relied on the opinion of Dr. Bercik, who concluded that Schink’s alleged mental impairments were not severe. The ALJ noted that although Dr. Bercik indicated that Schink had mood disorder, bipolar disorder, and personality disorder, she opined that his impairments caused him only mild restrictions in his activities of daily living, mild difficulties in maintaining social functioning, and mild difficulties in maintaining concentration, persistence, or pace. In the end, the ALJ accorded significant weight to the opinions of Drs. Bercik and Bernard, and minimal weight to the opinions of treating physicians Drs. Hernandez and Assad.

In determining that Schink’s bipolar disorder was a non-severe impairment, the ALJ concluded that Schink had only mild limitation in the area of activities of daily living because he could clean, shop, cook, pay bills, maintain a residence, and care appropriately for his personal hygiene, and he took care of two parrots. In addition, the ALJ determined that Schink had only mild limitation in social functioning because he was able to get along with others, such as family, friends, and neighbors, and he occasionally went to church. Finally, the ALJ found that Schink had only mild limitation in the area of concentration, persistence, and pace. The ALJ reasoned that because, in the ALJ’s view, Schink’s mental impairment caused no more than "mild" limitation in any of these functional areas, and because Schink had no episodes of decompensation of extended duration, his mental impairments were not severe.

Then the ALJ proceeded to step three of the sequential analysis because he had found that some of Schink’s physical impairments were severe. At this step, the ALJ determined that Schink did not have an impairment or combination of impairments that met or medically equaled any listed impairment. At step four, the ALJ found that through the date of last insured Schink "had the residual functional capacity to perform a full range of light work limited to lifting/carrying up to 20 pounds occasionally and up to 10 pounds frequently," could "sit, stand and/or walk for a total of six hours during an eight-hour workday," could "occasionally stoop or crouch," and could "frequently reach in all directions, including overhead." The ALJ did not include any mental capacities or limitations in the assessment of Schink’s residual functional capacity. In arriving at this ruling, the ALJ stated that he considered "all symptoms." The ALJ further narrated, but did not discuss, Schink’s testimony that he had previously received short-term disability benefits based on his bipolar-disorder diagnosis, that he could no longer perform work as a car salesman because he had become argumentative with managers and customers, and that he was prescribed medication for bipolar disorder.

Ultimately, the ALJ concluded that Schink could perform his past relevant work as a car salesman. In the alternative, the ALJ stated without explanation that even if Schink could not perform his past relevant work, other jobs existed in the national economy that he could perform.

Schink sought review of the denial of benefits by the Appeals Council. Among other things, Schink argued that the ALJ erred by failing to properly weigh and analyze the treating physicians’ opinions and by finding Schink’s mental impairments to be non-severe. Schink also argued for the first time that the ALJ was biased against him. In support of this contention, Schink alleged that the ALJ had been disciplined as a result of complaints filed against him by Schink’s counsel. Under the circumstances, Schink claimed that the ALJ should have recused himself from the case. The Appeals Council denied Schink’s request for review.

Schink later filed a complaint with the district court seeking review of the determination that he was not entitled to disability benefits. The district court affirmed the Commissioner’s decision to deny Schink disability benefits. The district court also rejected Schink’s bias claim. Schink timely appealed.

II. Standard of Review

We review this Social Security appeal to determine whether the Commissioner’s decision is supported by substantial evidence and whether the correct legal standards were applied. Winschel v. Comm’r of Soc. Sec. , 631 F.3d 1176, 1178 (11th Cir. 2011). When the Appeals Council denies review of the ALJ’s decision, as occurred here, we review the ALJ’s decision as the Commissioner’s final decision. Doughty v. Apfel , 245 F.3d 1274, 1278 (11th Cir. 2001).

Substantial evidence is "such relevant evidence as a reasonable person would accept as adequate to support a conclusion." Id. (citation omitted). Under this standard, we will not "decide the facts anew, reweigh the evidence, or substitute our judgment for that of the [Commissioner]." Id. (quoting Phillips v. Barnhart, 357 F.3d 1232, 1240 n. 8 (11th Cir. 2004) (alteration in original) (quoting Bloodsworth v. Heckler, 703 F.2d 1233, 1239 (11th Cir. 1983) ). But nor will we merely rubber-stamp a decision. We "must scrutinize the record as a whole to determine if the decision reached is reasonable and supported by substantial evidence." MacGregor v. Bowen , 786 F.2d 1050, 1053 (11th Cir. 1986).

III. Schink’s Bias Claim

Before turning to the merits of the appeal, we address Schink’s contention that the ALJ exhibited bias and should have recused himself. In the past, we have noted that "[t]he ALJ plays a crucial role in the disability review process. Not only is he duty-bound to develop a full and fair record, he must carefully weigh the evidence, giving individualized consideration to each claim that comes before him." Miles v. Chater , 84 F.3d 1397, 1401 (11th Cir. 1996) (per curiam). When the process is compromised, the claimant is entitled to an unbiased reconsideration of his application for benefits before a different ALJ. Id. ; see also 20 C.F.R. § 404.940.

The regulations themselves provide a process for disqualification, stating that an ALJ "shall not conduct a hearing if he or she is prejudiced or partial with respect to any party or has any interest in the matter pending for decision." 20 C.F.R. § 404.940. A claimant may object to the ALJ who is designated to conduct the hearing, but he must give notice to the ALJ of his objections at the "earliest opportunity." Id. The ALJ must then consider the claimant’s objections and decide whether to proceed or withdraw. Id. If the ALJ does not withdraw, the claimant may, after the hearing, present objections to the Appeals Council as to why a hearing decision should be changed or a new hearing held before another ALJ. Id.

The bulk of Schink’s bias claim stems from his contention that the ALJ harbored animus against his attorney. According to Schink, the animus is evident from a lawsuit the ALJ filed against the Commissioner. Schink claims, in that lawsuit, among other things, the ALJ accused Schink’s attorney of deceptive and fraudulent behavior. The ALJ alleged he noticed a "pattern" that had developed among several local attorneys who routinely requested interpreters in an attempt to bolster the illegitimate contention that the claimants could not "communicate in English." See Butler v. Colvin , No. 14-60444-cv-Williams/Turnoff (S.D. Fla. 2014); Doc. 1 at ¶ 13; Doc 1-2 at 43-44. The ALJ further alleged he was issued a reprimand because he declined to reschedule three cases for hearing using a Spanish interpreter in which Schink’s attorney represented claimants. Id. at Doc. 1 at ¶ 12; Doc 1-2 at 43. Based on this, Schink contends the ALJ was "embroiled in a personal dispute with Schink’s counsel and should have disqualified himself from the case."

We express no judgment about the merits of this contention because Schink did not raise the bias claim in a timely manner. The ALJ filed his lawsuit against the Commissioner—which, according to Schink, showed animus toward Schink’s counsel—on February 21, 2014. See Butler , No. 14-60444-cv-Williams/Turnoff (S.D. Fla. 2014); Doc. 1. Approximately one year and one month later, on March 16, 2015, the ALJ denied Schink’s claim for Social Security benefits. Schink had not raised any issue of alleged bias at that point and instead raised the issue of the ALJ’s alleged bias for the first time on April 17, 2015, when he appealed the ALJ’s denial of benefits. The failure to raise the bias claim earlier might be forgiven if Schink’s counsel had been unaware of the lawsuit. But here, Schink’s counsel knew about the lawsuit before the ALJ issued his March 16, 2015, decision and failed to raise the bias claim at the "earliest opportunity." He has therefore forfeited the claim. See 20 C.F.R. § 404.940 ; see also McKinney v. Pate , 20 F.3d 1550, 1562 (11th Cir. 1994) (explaining that "a challenger must object to a biased [district] judge in a motion to recuse before trial or as soon as the alleged bias is discovered," on pain of forfeiture). In particular, citations in Schink’s Opening Brief reveal this to be the case. In an attempt to convince this Court of the ALJ’s bias, Schink’s counsel pointed to several "factually similar" cases in which he represented clients who also had claims before the same ALJ and where the district court found a "risk of appearance of bias." When we reviewed those cases, it became obvious that counsel for Schink was aware of the ALJ’s complaint against the Commissioner by, at the latest, October 27, 2014. See King v. Comm’r of Soc. Sec. , Case No. 2:14-cv-341-CM (M.D. Fla. 2014); Doc. 21. On that date, in another case, counsel for Schink filed a Memorandum in Opposition to the Commissioner’s Decision in which he attached as an exhibit the ALJ’s complaint against the Commissioner. Id. Yet Schink did not allege any claim of bias in the pending matter until April 17, 2015—nearly six months later. Because Schink did not object to the ALJ’s alleged bias at the "earliest opportunity," Schink has forfeited the opportunity to complain of it now, and we affirm the district court’s order to the extent that it rejects his bias claim.

IV. Schink’s Substantive Claims

A. Treating Physicians’ Opinions

Much of Schink’s appeal centers on his contention that the ALJ improperly discounted the opinions of his treating physicians (Drs. Hernandez and Assad), who found Schink’s mental impairments to be severe and disabling. In Social Security cases, the opinions of a treating physician are entitled to more weight than those of a consulting or evaluating health professional. This is because treating physicians are more likely to be able to give a more complete picture of the applicant’s health history. As the Social Security Administration has explained, treating physicians

are likely to be the medical professionals most able to provide a detailed, longitudinal picture of [a claimant's] medical impairment(s) and may bring a unique perspective to the medical evidence that cannot be obtained from the objective medical findings alone or from reports of individual examinations, such as consultative examinations or brief hospitalizations.

20 C.F.R. § 404.1527(c)(2).

The ALJ must give a treating physician’s opinion "substantial or considerable weight unless good cause is shown to the contrary." Phillips , 357 F.3d at 1240 (citation omitted); see also 20 C.F.R. § 404.1527(c)(2). Good cause exists when (1) the treating physician’s opinion was not bolstered by the evidence, (2) the evidence supported a contrary finding, or (3) the treating physician’s opinion was conclusory or inconsistent with his or her own medical records. Winschel , 631 F.3d at 1179 ; Phillips , 357 F.3d at 1240-41. We have explained that the ALJ must clearly articulate the reasons for giving less weight to the opinion of a treating physician. Winschel , 631 F.3d at 1179 ; see also 20 C.F.R. § 404.1527(c)(2) (noting that "good reasons" must be provided in the decision for the weight given to treating source’s medical opinion). The failure to do so is reversible error. Lewis v. Callahan , 125 F.3d 1436, 1440 (11th Cir. 1997). The ALJ must consider many factors when weighing a medical opinion. See 20 C.F.R. § 404.1527(c). For instance, the Social Security regulations command that the ALJ consider (1) the examining relationship; (2) the treatment relationship, including the length and nature of the treatment relationship; (3) whether the medical opinion is amply supported by relevant evidence; (4) whether an opinion is consistent with the record as a whole; and (5) the doctor’s specialization. Id. Non-examining physicians’ opinions are entitled to little weight when they contradict opinions of examining physicians and do not alone constitute substantial evidence. Sharfarz v. Bowen , 825 F.2d 278, 280 (11th Cir. 1987) (per curiam).

After considering the record and with the benefit of oral argument, we find that the ALJ failed to articulate good cause for discounting Dr. Hernandez’s and Dr. Assad’s opinions in favor of the non-examining consultative physician, Dr. Bercik. First, on this record, the ALJ should not have discounted the treating physicians’ opinions based on what he perceived to be "sporadic" treatment. True, an ALJ is justified in discounting a physician’s opinion when the doctor has seen the claimant only once; for the purposes of our caselaw, "one-time examiners" are not properly considered "treating physicians." McSwain v. Bowen , 814 F.2d 617, 619 (11th Cir. 1987) (per curiam); see also Crawford v. Comm’r of Soc. Sec. , 363 F.3d 1155, 1160 (11th Cir. 2004) (per curiam). We have also held that an ALJ was justified in discounting a treating physician’s opinion when the physician "saw [the claimant] twice and submitted only sketchy, conclusory notes." Hudson v. Heckler , 755 F.2d 781, 784 (11th Cir. 1985) (per curiam). But in that case, it was not the low number of examinations alone that provided good cause to discount the opinion; what mattered more was that we found the opinion "so brief and conclusory that it lack[ed] persuasive weight" and that it could not be said to be "[ ]substantiated by any clinical or laboratory findings." Id. (quoting Bloodsworth , 703 F.2d at 1240 ).

But here, both doctors administered significant treatment to Schink multiple times over the course of months before completing the questionnaires that contained their ultimate opinions on his mental impairments. Dr. Hernandez saw Schink at least three times over the course of five months before providing his opinion. He helped to manage Schink’s treatment plan, and he prescribed Schink medications and altered their doses based on Schink’s response. As for Dr. Assad, he saw Schink at least eight times before assessing Schink’s mental impairments as indicated on the questionnaire. He administered cognitive-behavioral therapy to Schink, as reflected in his detailed notes of their therapy sessions. Dr. Assad’s notes from Schink’s intake appointment also record that Dr. Assad coordinated his treatment with Dr. Hernandez, who had referred Schink to him in the first place. For these reasons, the ALJ’s decision to discount Dr. Hernandez’s and Dr. Assad’s opinions was not supported by the suggestion that Schink saw them only infrequently. Both were undoubtedly treating physicians, and their familiarity with Schink was sufficient to entitle their opinions to the presumption of substantial or considerable weight that is ordinarily due to treating physicians’ opinions.

What is more, the ALJ gave "significant weight" to the opinions of Drs. Bernard and Bercik, even though Dr. Bernard saw Schink only once and Dr. Bercik never saw him at all. The ALJ’s failure to apply his sporadic-treatment rationale across the board—with no explanation given and with no obvious reason for the inconsistency in sight—makes it impossible for us to consider this rationale "good cause." See Lewis , 125 F.3d at 1440–41 (rejecting the ALJ’s decision to discount a treating physician’s opinion because he was not a cardiologist when the ALJ did not apply the same criterion to non-treating physicians); cf. Spencer by Spencer v. Heckler , 765 F.2d 1090, 1094 (11th Cir. 1985) (per curiam) (rejecting the ALJ’s inconsistent reliance on one source’s opinion to undermine that of a treating physician when the ALJ rejected the first source’s opinion about other important matters on unexplained grounds).

Next, the ALJ improperly rejected the opinions of the treating physicians based on the format of the questionnaires completed by Drs. Hernandez and Assad. The ALJ objected that the questionnaires used vague language, failed to track the language of the regulatory regime and official forms used by the Social Security Administration, and failed to "address the category of ‘Understanding and Memory’ at all." None of these reasons amounts to good cause for discounting the questionnaires.

First, the regulations do not require a doctor’s opinion to take a certain form. On the contrary, they expressly contemplate that medical sources "may"—but need not—use terms similar to those used in the regulations and may—but need not—use them in exactly the same way as the Administration if they do so. See 20 C.F.R. pt. 404, subpt. P., app. 1, § 12.00(F)(3)(a) ("The medical evidence may include descriptions regarding the diagnostic stage or level of [a claimant’s] disorder, such as ‘mild’ or ‘moderate.’ Clinicians may use these terms to characterize [a] medical condition. However, these terms will not always be the same as the degree of [a claimant’s] limitation in a paragraph B area of mental functioning."). Where no other evidence counters the treating physician’s opinion, the ALJ cannot reject that opinion simply because it is not in a particular format. Instead, the ALJ must address the merits of the treating physicians’ opinions. Here, although the ALJ found certain terms, like "mild" and "extreme," to be vague, the questionnaire defined those terms. Consequently, how the treating doctors assessed Schink’s mental impairments should not have been a mystery to the ALJ. As for the phrase "unable to function," it has a commonsense meaning and should not have been objectionable to the ALJ. The same is true of the ALJ’s suggestion that time periods like "50% of the work day or work week" were difficult to understand.

Plus, the ALJ’s conclusion that he found the terms Dr. Hernandez’s and Dr. Assad’s opinions employed to be vague is contradicted by the fact that the state consultative doctor who opined about Schink’s condition—Dr. Bercik—used the same terms, and the ALJ had no problem relying on Dr. Bercik’s opinion. Indeed, when determining Schink’s "degree of limitation," Dr. Bercik used a check-box form that similarly employed the terms "mild" and "extreme." The ALJ’s reliance on Dr. Bercik’s opinion therefore negates this rationale for discounting the treating doctors’ questionnaires. See Lewis , 125 F.3d at 1440–41. As for the ALJ’s objection that the questionnaire failed to address Schink’s understanding and memory, we think it is beside the point. Schink’s claim is that his bipolar disorder disables him primarily by affecting his mood, affect, and interpersonal relationships—not his cognition or memory. Cf. MacGregor , 786 F.2d at 1053 (holding that the claimant’s lack of intellectual impairment was not a sufficient reason to discount a treating physician’s opinion of his mood impairment). To be sure, there is considerable evidence that Schink’s conditions affect his cognition and memory. But the heart of his claim lies elsewhere. And in any event, a medical opinion’s failure to address all possible functional limitations is not a logical reason to discount what it says about the limitations that it does address.

The most that can be said in criticism of the questionnaires’ format is that they used a "check box" format with limited space for explanation of the assessments. But that is not a basis, in and of itself, to discount them as conclusory. For one thing, the same was true of Dr. Bercik’s opinion, which the ALJ relied on heavily, and as we have explained, a rationale applied inconsistently for no apparent reason is not good cause. More importantly, treating-physician opinions "should not be considered in a vacuum, and instead, the doctors’ earlier reports should be considered as the bases for their statements." Wilson v. Heckler , 734 F.2d 513, 518 (11th Cir. 1984) (per curiam). In other words, the ALJ should have interpreted Drs. Hernandez’s and Assad’s answers to the questionnaires in light of their treatment notes.

Here, the doctors’ treatment notes fleshed out and were consistent with their conclusions regarding Schink’s mental health as set forth on the questionnaires. Dr. Hernandez’s notes reflected that Schink repeatedly presented with racing thoughts, depression with dysthymic and anhedonic characteristics, anxiety, and a "labile"—that is, unstable or changeable—affect, and that he had a family history of suicide. Dr. Hernandez also diagnosed Schink with bipolar disorder and anxiety disorder. As for Dr. Assad, he repeatedly found Schink to have pressured speech, impaired judgment, elevated anxiety and depression, and hypomanic affect. Dr. Assad similarly diagnosed Schink with bipolar disorder and depression, and his notes from before filling out the questionnaire reflect his familiarity with Schink’s interpersonal and emotional difficulties, including a "repetitive pattern in most relationships" of "intense anger" and passive aggression. At one of his sessions with Schink, Dr. Assad wrote, "clearly bipolar lability is continuing." Without question, the treatment notes supported the questionnaires filled out by both doctors.

The ALJ found that Drs. Hernandez’s and Assad’s notes "indicate[d] only mild limitations in reported mental status examinations, at best," but the ALJ did not "clearly articulate" the basis for this conclusion, Lewis , 125 F.3d at 1440, nor do we see how the record could support it. To be sure, some of Schink’s mental-status examinations were better than others, and at each visit he appeared better on some parameters than on others. For instance, as the ALJ narrated in the background section of his opinion, Dr. Assad recorded at one therapy session that Schink displayed "tangential" thought processes but "was able to be redirected and remain on topic," and Dr. Hernandez recorded at his first appointment with Schink that Schink was "cooperative" and exhibited "organized" speech, "relevant" thought content, "fair" insight, and "intact" cognition. But to discount a treating physician’s opinion because it is "inconsistent with [the source’s] own medical records," an ALJ must identify a genuine "inconsisten[cy]." Lewis , 125 F.3d at 1440. It is not enough merely to point to positive or neutral observations that create, at most, a trivial and indirect tension with the treating physician’s opinion by proving no more than that the claimant’s impairments are not all-encompassing. See MacGregor , 786 F.2d at 1053–54 (explaining that there was "no inconsistency whatever" between a treating physician’s conclusion that the claimant was so depressed "that he could not operate under pressure nor relate appropriately to supervisors or co-workers" and the same doctor’s statement that the claimant was "intelligent enough to understand and follow orders and to solve problems"; after all, "highly intelligent and able people do fall prey to crippling depression"). And the ALJ’s opinion does not so much as hint at any real inconsistency between Drs. Assad’s and Hernandez’s treatment notes and their opinions in the questionnaires. For example, it is not inconsistent—or even that unlikely—that a patient with a highly disruptive mood disorder, in a structured one-on-one conversation with a mental-health professional, might be capable of "be[ing] redirected" from his "tangential" thought processes so as to "remain on topic."

Nor can we accept the ALJ’s finding that Drs. Hernandez’s and Assad’s opinions in the questionnaires were "inconsistent with other substantial evidence of record" as a good reason for discounting them, for two reasons. First, once again, the ALJ failed to clearly articulate what evidence led him to this conclusion. See Lewis , 125 F.3d at 1440 ; see also Winschel , 631 F.3d at 1179 ("[T]he ALJ must state with particularity the weight given to different medical opinions and the reasons therefor ." (emphases added)); MacGregor , 786 F.2d at 1053 ("The [ALJ] must specify what weight is given to a treating physician’s opinion and any reason for giving it no weight ...." (emphases added)). Second, once again, we fail to see the inconsistency.

Indeed, the record as a whole strikes us as consistent with the treating physicians’ opinions. For example, the opinions in the questionnaires comported with Dr. Anthony’s assessment that Schink’s affect was "blunted," his energy was low, and he had "marginal social and interpersonal involvement." And the treating physicians’ opinions about Schink’s social functionality are consistent with Dr. Bernard’s evaluation, to which the ALJ "accorded significant weight." Dr. Bernard diagnosed Schink with a mood disorder, a personality disorder with cluster B features, "[p]roblems dealing with the social environment," and "occupational limitations." He recorded that Schink’s affect was irritable; that his attitude was "brusque, arrogant, flippant, and abrasive"; and that "[o]verall, he had a harshness and domineering aspect to his personality." On their face, these observations by Dr. Bernard are consistent with the treating physicians’ opinions that Schink’s mental-health conditions would substantially impair his social interactions and emotional resilience in the workplace. If the ALJ discounted these observations despite ostensibly relying on Dr. Bernard’s examination in other respects, he should have explained why. Cf. Spencer , 765 F.2d at 1094. And if the ALJ gave "significant weight" to these observations along with the rest of Dr. Bernard’s report, but nonetheless concluded that the rest of the evidence was inconsistent with the treating physicians’ opinions in the questionnaires, he should have given his reasons for reaching that far-from-obvious conclusion.

We recognize that the ALJ expressed his belief that Schink "was able to participate in normal activities of daily living." But the daily activities upon which the ALJ relied were mostly, if not all, solitary activities such as watching television, walking the dog, and cooking. These activities do not discount the treating physicians’ opinions that Schink suffered significantly from mental impairments, particularly when he interacted with others.

Finally, we reject the government’s suggestion that we affirm based on the ALJ’s statement that Drs. Hernandez and Assad "did not cite to any medically acceptable clinical or diagnostic techniques to support their opinions." Even if the discounting of their opinions could have been justified on this basis with a proper explanation—a matter we do not consider—the ALJ provided no explanation for this statement, leaving it an unadorned echo of a legal standard from the regulations. See 20 C.F.R. § 404.1527(c)(2). So it cannot discharge the ALJ’s duty to "clearly articulate the reasons for giving less weight to the opinion of a treating physician." Lewis , 125 F.3d at 1440.

For all these reasons, we conclude that the ALJ failed to articulate good cause for discounting the opinions of Drs. Hernandez and Assad.

B. Severity of Schink’s Mental Impairments

The ALJ ultimately denied Schink’s disability claim because he found that Schink did not suffer from a severe mental impairment and could return to his past job as a car salesman. We agree with Schink that substantial evidence did not support the ALJ’s finding that Schink’s mental impairments—most notably his bipolar disorder—were non-severe as defined by 20 C.F.R. § 404.1521 We also conclude that the ALJ erred by failing to consider Schink’s mental capabilities and limitations when he conducted his Residual Functional Capacity ("RFC") assessment.

The Social Security regulations set forth a five-step, sequential evaluation process to determine whether a claimant is disabled. At the second step of the sequential evaluation, the ALJ must "consider the medical severity of [the claimant’s] impairment(s)." Phillips , 357 F.3d at 1237 (alteration in original) (quoting 20 C.F.R. § 404.1520(a)(4)(ii) ). This step is a "threshold inquiry" and "allows only claims based on the most trivial impairments to be rejected." McDaniel v. Bowen , 800 F.2d 1026, 1031 (11th Cir. 1986). The burden rests with the claimant, however, to show that he has a severe impairment or combination of impairments. Jones v. Apfel , 190 F.3d 1224, 1228 (11th Cir. 1999).

"An impairment or combination of impairments is not severe if it does not significantly limit [the claimant’s] physical or mental ability to do basic work activities." 20 C.F.R. § 404.1522(a) ; see also Phillips , 357 F.3d at 1237. Basic work activities include the following:

(1) Physical functions such as walking, standing, sitting, lifting, pushing, pulling, reaching, carrying, or handling; (2) Capacities for seeing, hearing, and speaking; (3) Understanding, carrying out, and remembering simple instructions; (4) Use of judgment; (5) Responding appropriately to supervision, co-workers and usual work situations ; and (6) Dealing with changes in a routine work setting.

20 C.F.R. § 404.1522(b) (emphasis added).

We have recognized that an "impairment is not severe only if the abnormality is so slight and its effect so minimal that it would clearly not be expected to interfere with the individual’s ability to work, irrespective of age, education or work experience." McDaniel , 800 F.2d at 1031. A claimant’s burden to establish a severe impairment at step two is only "mild." Id.

Based on these standards, substantial evidence does not support the ALJ’s conclusion that Schink’s mental impairments were not severe. On this record, Schink’s impairments due to his bipolar disorder, anxiety, and mood disorder cannot be considered only "slight" or "trivial" abnormalities. Schink’s mental-health issues were serious enough that he was referred to and saw various mental-health professionals over a period of years. Every doctor who saw Schink diagnosed him with bipolar disorder or a comparable personality disorder and opined that it significantly affected his mood, affect, and ability to interact with others. No state doctor disputed this diagnosis. On the contrary, the only state doctor who examined Schink, Dr. Bernard, attested to his "brusque, arrogant, flippant, and abrasive" attitude, his "harshness and domineering aspect," and his "irritable" affect, among other traits corroborating a non-trivial personality disorder. The evidence bears out these remarks by Dr. Bernard. Here, the evidence showed that Schink was argumentative and combative with others, regularly harbored revenge fantasies, and even described wanting to kill his neighbor. Dr. Assad’s psychotherapy notes record many other instances of Schink’s impulsive and irascible tendencies spurring him into conflict with the people around him. Anger, mania, depression, and conflicted interpersonal relationships were present in Schink’s symptomatology and surely would have had some effect on Schink’s ability to respond to supervision and coworkers.

Further, Schink’s GAF scores did not support a finding that Schink’s mental impairments should be considered "slight." The known GAF scores ranged from 50 to 60, reinforcing that Schink had at least moderate difficulty in functioning. Even the state consultative examiner Dr. Bernard, who assessed Schink for his ability to work, determined a GAF score of 59. We recognize that GAF scores are by no means dispositive of a claim, but, in this case, Schink’s GAF scores and the more detailed medical evidence point to the same conclusion: that his mental-health condition constitutes a severe impairment.

The ALJ based his finding of non-severity in part on the conclusion that Schink had only a "mild" limitation in his activities of daily living, stating that he could clean, shop, cook, pay bills, maintain a residence, and care for his own grooming needs. But that conclusion was not substantially supported by the evidence. Dr. Bernard recorded that Schink "attempt[ed] to cook only minimally" and "[did] very little housework." These assessments were consistent with Schink’s testimony before the ALJ and with a June 2010 function report in which Schink wrote that he cooked "easy stuff mostly," like "sandwiches" and "frozen dinners," and that he did not "clean much," resulting in "stuff pil[ing] up." And while Schink took care of his grooming needs and could pay bills, this hardly constitutes a full range of daily activities, and it hardly establishes that Schink’s mental-health issues were "so slight and [their] effect so minimal that [they] would clearly not be expected to interfere with [his] ability to work" in any significant way. McDaniel , 800 F.2d at 1031.

Indeed, Social Security regulations acknowledge that the ability to complete tasks in settings that are less demanding than a typical work setting "does not necessarily demonstrate [an applicant’s] ability to complete tasks in the context of regular employment during a normal workday or work week." 20 C.F.R. pt. 404, subpt. P, app. 1, § 12.00(C)(6)(b). That is especially relevant where, as here, an applicant spends most of his time among familiar (or no) people and a steady environment, so his behavior does not necessarily show how he would function in a work setting on a sustained basis. See id. pt. 404, subpt. P, app. 1, § 12.00(D)(3)(b). So that Schink was able to feed and clothe himself, walk a dog, and watch television tells us very little about whether he suffered from a severe mental impairment or about his ability to function in a stressful work setting.

The ALJ also found that Schink had only mild limitations in social functioning because he could interact independently, appropriately, and effectively on a sustained basis with other individuals. Again, no support for this conclusion exists in the record. Rather, the record shows that Schink leads an isolated life, rarely engaging in activities outside the home, with few or no friends, and with major and chronic conflict in his few significant relationships. In fact, the ALJ acknowledged that Schink spent most of his day watching television, playing on the computer, napping, and going for long drives. These activities—which do not require or even involve human interaction—do not establish that Schink is able to function socially. Instead, the record painted a picture of a depressed, agitated, frequently angry, and sometimes tearful person who had a family history of mental illness and who had for years seen doctors and taken medication to control his disruptive bipolar disorder.

The ALJ found that medication helped to manage Schink’s symptoms and that his "on-going treatment of medication management and therapy (counseling) has resulted in ... a level of adaptation adequate for employment on a regular basis." Whether or not Schink’s impairments, when treated, would be compatible with employment—the question reserved for steps four and five of the sequential analysis—the effects of treatment on Schink are not substantial evidence that his mental impairments were non-severe. Schink’s treatment with medication was intermittent, his prescriptions changed frequently, and he complained that antidepressants "made him worse" and a mood stabilizer "made him feel weird." True, Schink sometimes spoke positively about his medications; notably, he told Dr. Assad days after his discharge from the hospital that his then-current prescriptions were helping. (Over the previous fortnight, Schink had been prescribed at least five different psychotropic medications to control his near-suicidal state.) But a mere month later, Schink testified to the ALJ that his doctor was experimenting with "different medications," and his mental condition was obviously poor. Considering the record as a whole, there is little evidence that medication has ever durably improved Schink’s condition, and there certainly is no substantial evidence that it has ever made it "so slight and its effect so minimal that it would clearly not be expected to interfere with [his] ability to work." McDaniel , 800 F.2d at 1031. Similarly, even if psychotherapy has helped Schink to a certain extent, it is obvious that he still suffers from serious emotional and interpersonal challenges even after years of counseling sessions with Dr. Assad.

Nor does the fact that Schink, at times, seemed to be "doing better" support a finding on this record that Schink’s mental impairments were non-severe. Indeed, the bulk of the treatment notes support the notion that Schink’s mental impairments continued well beyond his brief periods of stability. In this respect, the treatment notes reflect the episodic nature of bipolar disorder and refute the lack of a severe mental impairment.

We agree with our sister Circuits that people with chronic diseases can experience good and bad days. And when bad days are extremely bad and occur with some frequency, they can severely affect a person’s ability to work:

A person who has a chronic disease, whether physical or psychiatric, and is under continuous treatment for it with heavy drugs, is likely to have better days and worse days; that is true of the plaintiff in this case. Suppose that half the time she is well enough that she could work, and half the time she is not. Then she could not hold down a full-time job. That is likely to be the situation of a person who has bipolar disorder that responds erratically to treatment.

Bauer v. Astrue, 532 F.3d 606, 609 (7th Cir. 2008) ; accord Singletary v. Bowen, 798 F.2d 818, 821 (5th Cir.1986) (noting that although "symptom-free periods may negate a finding of disability when a physical impairment is alleged, symptom-free intervals do not necessarily compel such a finding when a mental disorder is the basis of the claim" (citation omitted)).

Here, that Schink experiences good days and bad days is to be expected. On this record, and considering the episodic nature of Schink’s mental impairment, the ALJ’s citation of the good days as evidence of no disability did not support a finding that Schink did not suffer from a severe impairment (or that his doctors’ treatment opinions are inconsistent with the record). Indeed, even Dr. Bernard (the state consultative doctor) found Schink’s prognosis to be "chronic." That Schink had to be hospitalized for one week due to intense suicidal thoughts and depression is revealing about the depth of Schink’s condition.

Taking everything together, we cannot conclude that substantial evidence supported the ALJ’s finding that Schink’s mental impairments were non-severe. On this record, it cannot be said that Schink’s bipolar disorder and mood disorder were abnormalities so slight and trivial that they would produce minimal effects on Schink’s ability to work.

C. The ALJ’s Deficient RFC Assessment

Our conclusion that substantial evidence does not support the ALJ’s finding that Schink’s mental impairments were non-severe does not necessarily end the discussion. That finding could be harmless if the ALJ nevertheless proceeded in the sequential evaluation, duly considered Schink’s mental impairment when assessing his RFC, and reached conclusions about Schink’s mental capabilities supported by substantial evidence. Here, though, the ALJ’s RFC assessment was limited to Schink’s physical abilities and impairments and erroneously omitted his mental ones. As a result, we cannot say that the erroneous finding of non-severity was harmless.

At step four of the sequential analysis, the ALJ conducts a residual-functional-capacity assessment of the claimant, which is "an assessment, based upon all of the relevant evidence, of a claimant’s remaining ability to do work despite his impairments." Lewis , 125 F.3d at 1440 (citing 20 C.F.R. § 404.1545(a) ). The ALJ makes this determination by considering a claimant’s physical, mental, and other abilities affected by the impairment. See 20 C.F.R. § 404.1545(b) - (d). A limited ability to carry out certain mental activities, such as limitations which affect "responding appropriately to supervision, co-workers, and work pressures in a work setting," may reduce a claimant’s ability to do past work and other work. Id. § 404.1545(c).

To support his conclusion that Schink was able to return to his past job as a car salesman, the ALJ was required to consider all the duties of that work and evaluate Schink’s ability to perform them despite his impairments. Lucas v. Sullivan , 918 F.2d 1567, 1574 (11th Cir. 1990). Consideration of all impairments, severe and non-severe, is required when assessing a claimant’s RFC. Bowen v. Heckler , 748 F.2d 629, 634-35 (11th Cir. 1984). The ALJ must also consider a claimant’s medical condition taken as a whole. Mitchell v. Comm’r, Soc. Sec. Admin. , 771 F.3d 780, 782 (11th Cir. 2014) ; Phillips , 357 F.3d at 1237 (ALJ has a duty to consider impairments in combination and to determine whether combined impairments render the claimant disabled); see also 20 C.F.R. § 404.1523(c) and Social Security Ruling 96-8p. If an ALJ fails to address the degree of impairment caused by the combination of physical and mental medical problems, the decision that the claimant is not disabled cannot be upheld. Bowen , 748 F.2d at 634 ("[I]t is certain that mental and psychological defects can combine with physical impairments to create total disability to perform gainful employment." (quoting Brenem v. Harris, 621 F.2d 688, 690 (5th Cir.1980) )).

Here, although the ALJ stated he "considered all symptoms" when assessing Schink’s RFC, the content of his decision demonstrates he did not. Nearly the entire section of the ALJ’s opinion relating to RFC discusses Schink’s physical impairments. For instance, the decision discusses at length Schink’s obesity, diabetes, right shoulder problems, knee pain, and sleep apnea. And while it mentions that Schink had bipolar disorder, the decision contains no real discussion of how the mental condition affected Schink’s RFC. Indeed, most of the references to Schink’s bipolar disorder in the RFC section are purely biographical or occur within summaries of medical examinations relating to Schink’s physical conditions. Cf. Ambers v. Heckler , 736 F.2d 1467, 1470 (11th Cir. 1984) ("[I]t does not appear that the ALJ considered Ambers’ other psychological impairments. ... The ALJ made no findings on these, other than to restate the physicians’ diagnoses of these impairments."). In fact, the ALJ’s ultimate conclusions as to RFC do not include even a single finding about Schink’s mental capacities. Instead, the ALJ’s findings concern Schink’s physical capacities exclusively.

Even the most favorable interpretation of the ALJ’s opinion—namely, that the ALJ considered Schink’s mental conditions in the RFC assessment sub silentio and implicitly found that they imposed no significant limitations on his work-related mental capacities—would not permit us to affirm because, as our precedent holds, the ALJ’s "failure ... to provide the reviewing court with sufficient reasoning for determining that the proper legal analysis has been conducted mandates reversal" in its own right. Keeton v. Dep’t of Health & Human Servs. , 21 F.3d 1064, 1066 (11th Cir. 1994). We recognize that in finding Schink’s bipolar disorder to be a non-severe impairment, the ALJ went through the four broad functional areas known as the "paragraph B" criteria. But the ALJ also explained that the "limitations identified in the ‘paragraph B’ criteria are not a residual functional capacity assessment but are used to rate the severity of mental impairments at steps 2 and 3 of the sequential evaluation process." As acknowledged by the ALJ in his opinion, the mental RFC assessment used at steps 4 and 5 of the process "requires a more detailed assessment by itemizing various functions contained in the broad categories found in paragraph B of the adult mental disorders listings in 12.00 of the Listing of Impairments." Even if we assume the RFC assessment conducted by the ALJ included some silent consideration of Schink’s mental impairments, we have no way of knowing whether it included the "more detailed assessment" required.

Severe or not, the ALJ was required to consider Schink’s mental impairments in the RFC assessment but evidently failed to do so. And as a result of this error, we cannot say that the ALJ’s earlier error in finding Schink’s mental impairments to be non-severe was harmless. In short, the ALJ provided no real assessment of how Schink’s mental impairments—including depression, mania, and anger—affected his ability to work. The assessment was therefore inadequate.

V. Conclusion

For the foregoing reasons, we find that Schink’s claim of bias was forfeited but that the ALJ failed to articulate good cause for discounting the opinions of Drs. Hernandez and Assad, the ALJ’s finding of non-severity is not supported by substantial evidence, and the ALJ failed to consider Schink’s mental impairments in assessing his RFC. Accordingly, the judgment of the district court is affirmed in part and reversed in part, and this case is remanded to the district court with instructions to vacate the Commissioner’s decision and to remand to the Commissioner for further proceedings consistent with this opinion.

AFFIRMED IN PART, REVERSED IN PART, AND REMANDED WITH INSTRUCTIONS.